{"schemaVersion":1,"title":"Bright Starter Shelf","revision":"2026-10-12-r1","reviewedAt":"2026-10-12","canonicalUrl":"https://brightaifuture.com/open-models#starter-shelf","methodology":"Explicit editorial starting options from documentation, not a size ranking or demonstrated hardware fit. All 48 bounded machine/task/runtime choices retain unsupported and needs-verification states. Download bytes are not runtime memory; RAM and dedicated VRAM are separate.","rights":"Source materials and runtime packages retain their existing licenses.","answers":[{"id":"mac-8gb-write-lm-studio","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-8gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/write","indexable":true,"candidateId":"smollm3_3b_q4","candidate":{"id":"smollm3_3b_q4","name":"SmolLM3 3B Q4_K_M","makerCheckpoint":"HuggingFaceTB/SmolLM3-3B","makerModelCardUrl":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","license":"Apache-2.0","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"jobs":["write","files"],"starterContextTokens":2048,"contextProvenance":"Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit.","selectionReason":"A small text-only starting option for a constrained device, linked to the existing setup guide. Its selection does not establish speed, quality or memory use.","runtimeEvidence":[{"id":"smollm3_3b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"smollm3_3b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"smollm3_3b_q4_lm-studio_metal","runtimeEvidence":{"id":"smollm3_3b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"SmolLM3 preserves the small text-only guide path for this constrained Mac. The short context and one chat keep the exercise bounded; this is an editorial starting option, not a measured memory guarantee.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":2048,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","LM Studio recommends 16 GB or more. This 8 GB option is constrained and conditional: close heavy apps, use 2048 context, and stop if memory pressure, swapping or a failed load occurs.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_maker","title":"SmolLM3 3B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_artifact","title":"SmolLM3 3B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/HuggingFaceTB/SmolLM3-3B","https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 2048 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Use the SmolLM3 system setting","text":"For this short exercise, put /no_think in the system prompt, as the SmolLM3 maker documents. Keep its supplied template. Do not apply this flag to other families; its operation in this exact app and file is untested.","sourceIds":["smollm3_3b_q4_maker"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: SmolLM3 3B Q4_K_M\nApple Silicon Mac, 8 GB unified memory\nRuntime: ggml-org/SmolLM3-3B-GGUF\nStarter context: 2048 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-8gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit."]},{"id":"mac-8gb-write-ollama","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"needs-verification","canonicalPath":"/run/mac-8gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/write?runtime=ollama","indexable":false,"candidateId":"smollm3_3b_q4","candidate":{"id":"smollm3_3b_q4","name":"SmolLM3 3B Q4_K_M","makerCheckpoint":"HuggingFaceTB/SmolLM3-3B","makerModelCardUrl":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","license":"Apache-2.0","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"jobs":["write","files"],"starterContextTokens":2048,"contextProvenance":"Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit.","selectionReason":"A small text-only starting option for a constrained device, linked to the existing setup guide. Its selection does not establish speed, quality or memory use.","runtimeEvidence":[{"id":"smollm3_3b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"smollm3_3b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"smollm3_3b_q4_ollama_metal","runtimeEvidence":{"id":"smollm3_3b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"The Smol Ollama identifier is a mutable Hugging Face import path; exact package/runtime support is not sufficiently established for a launch starting option. Use the reviewed LM Studio path.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":2048,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","LM Studio recommends 16 GB or more. This 8 GB option is constrained and conditional: close heavy apps, use 2048 context, and stop if memory pressure, swapping or a failed load occurs.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact","smollm3_3b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_maker","title":"SmolLM3 3B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_artifact","title":"SmolLM3 3B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_ollama","title":"SmolLM3 3B Q4_K_M: separately documented Ollama package","url":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/HuggingFaceTB/SmolLM3-3B","https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","https://huggingface.co/ggml-org/SmolLM3-3B-GGUF","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":null,"revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[],"copyText":"Write and summarize: SmolLM3 3B Q4_K_M\nApple Silicon Mac, 8 GB unified memory\nRuntime: hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M\nStarter context: 2048 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-8gb/write?runtime=ollama","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit."]},{"id":"mac-8gb-image-lm-studio","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"unsupported","canonicalPath":"/run/mac-8gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/image","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"No image starting option is reviewed for an 8 GB Mac. Use a reviewed higher-memory profile rather than treating a small download as evidence of image compatibility.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"mac-8gb-image-ollama","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"unsupported","canonicalPath":"/run/mac-8gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/image?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"No image starting option is reviewed for an 8 GB Mac. Use a reviewed higher-memory profile rather than treating a small download as evidence of image compatibility.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"mac-8gb-files-lm-studio","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-8gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/files","indexable":true,"candidateId":"smollm3_3b_q4","candidate":{"id":"smollm3_3b_q4","name":"SmolLM3 3B Q4_K_M","makerCheckpoint":"HuggingFaceTB/SmolLM3-3B","makerModelCardUrl":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","license":"Apache-2.0","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"jobs":["write","files"],"starterContextTokens":2048,"contextProvenance":"Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit.","selectionReason":"A small text-only starting option for a constrained device, linked to the existing setup guide. Its selection does not establish speed, quality or memory use.","runtimeEvidence":[{"id":"smollm3_3b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"smollm3_3b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"smollm3_3b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"hf.co/ggml-org/SmolLM3-3B-GGUF:Q4_K_M","publisher":"ggml-org via Hugging Face import","revision":null,"files":[],"transferBytes":null,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"This Hugging Face import tag still needs verification of the import steps and compatible model/runtime versions. No oldest compatible version is established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"taskSourceIds":["smollm3_3b_q4_maker","smollm3_3b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"smollm3_3b_q4_lm-studio_metal","runtimeEvidence":{"id":"smollm3_3b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"ggml-org/SmolLM3-3B-GGUF","publisher":"ggml-org","revision":"4965cb60b150737b68a0408c36aeefb65078f894","files":[{"name":"SmolLM3-Q4_K_M.gguf","role":"language-model","bytes":1915305312,"sha256":"8334b850b7bd46238c16b0c550df2138f0889bf433809008cc17a8b05761863e","listingUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/blob/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/ggml-org/SmolLM3-3B-GGUF/resolve/4965cb60b150737b68a0408c36aeefb65078f894/SmolLM3-Q4_K_M.gguf"}],"transferBytes":1915305312,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. Use current stable LM Studio with a compatible llama.cpp engine. The exact oldest compatible version for this file is not established.","licenseSourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact"],"taskSourceIds":["smollm3_3b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"SmolLM3 preserves the small text-only guide path for this constrained Mac. The short context and one chat keep the exercise bounded; this is an editorial starting option, not a measured memory guarantee.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":2048,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","LM Studio recommends 16 GB or more. This 8 GB option is constrained and conditional: close heavy apps, use 2048 context, and stop if memory pressure, swapping or a failed load occurs.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","smollm3_3b_q4_maker","smollm3_3b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_maker","title":"SmolLM3 3B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/HuggingFaceTB/SmolLM3-3B","checkedAt":"2026-10-12"},{"id":"smollm3_3b_q4_artifact","title":"SmolLM3 3B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/HuggingFaceTB/SmolLM3-3B","https://huggingface.co/api/models/ggml-org/SmolLM3-3B-GGUF/revision/4965cb60b150737b68a0408c36aeefb65078f894?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 2048 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Use the SmolLM3 system setting","text":"For this short exercise, put /no_think in the system prompt, as the SmolLM3 maker documents. Keep its supplied template. Do not apply this flag to other families; its operation in this exact app and file is untested.","sourceIds":["smollm3_3b_q4_maker"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: SmolLM3 3B Q4_K_M\nApple Silicon Mac, 8 GB unified memory\nRuntime: ggml-org/SmolLM3-3B-GGUF\nStarter context: 2048 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-8gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 2048 tokens for this constrained exercise. This supersedes longer context suggestions for the 8 GB Mac starting option. It is not the model’s maximum context or a tested memory limit.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"mac-8gb-files-ollama","machineId":"mac-8gb","machine":{"id":"mac-8gb","label":"Apple Silicon Mac, 8 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":8,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/mac-8gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-8gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"mac-16gb-write-lm-studio","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-16gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 16 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-16gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-16gb-write-ollama","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-16gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 16 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-16gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-16gb-image-lm-studio","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-16gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 16 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-16gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"mac-16gb-image-ollama","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-16gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 16 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-16gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"mac-16gb-files-lm-studio","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-16gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 16 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-16gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"mac-16gb-files-ollama","machineId":"mac-16gb","machine":{"id":"mac-16gb","label":"Apple Silicon Mac, 16 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":16,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/mac-16gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-16gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"mac-32gb-write-lm-studio","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-32gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 32 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-32gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-32gb-write-ollama","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-32gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 32 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-32gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-32gb-image-lm-studio","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-32gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 32 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-32gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"mac-32gb-image-ollama","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-32gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 32 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-32gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"mac-32gb-files-lm-studio","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-32gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 32 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-32gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"mac-32gb-files-ollama","machineId":"mac-32gb","machine":{"id":"mac-32gb","label":"Apple Silicon Mac, 32 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":32,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/mac-32gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-32gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"mac-64gb-write-lm-studio","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-64gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 64 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-64gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-64gb-write-ollama","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-64gb/write","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 64 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-64gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"mac-64gb-image-lm-studio","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-64gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 64 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-64gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"mac-64gb-image-ollama","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/mac-64gb/image","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_metal","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 64 GB unified memory\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-64gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"mac-64gb-files-lm-studio","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/mac-64gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_metal","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nApple Silicon Mac, 64 GB unified memory\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/mac-64gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"systemRamGB":null,"operatingSystem":"macOS 14 or newer","processor":"Apple Silicon M series; Intel Macs excluded","backend":"metal","driver":null,"starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","Apple unified memory is one shared pool, not separate CPU and GPU memory.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"mac-64gb-files-ollama","machineId":"mac-64gb","machine":{"id":"mac-64gb","label":"Apple Silicon Mac, 64 GB unified memory","platform":"apple-silicon","unifiedMemoryGB":64,"dedicatedVramGB":null,"minimumSystemRamGB":null},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/mac-64gb/files","canonicalUrl":"https://brightaifuture.com/run/mac-64gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"nvidia-8gb-write-lm-studio","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-8gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 8 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 8 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-8gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-8gb-write-ollama","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-8gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"systemRamGB":16,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 8 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 8 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-8gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-8gb-image-lm-studio","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-8gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 8 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 8 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-8gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"nvidia-8gb-image-ollama","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-8gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"systemRamGB":16,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 8 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 8 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-8gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"nvidia-8gb-files-lm-studio","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-8gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 8 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 8 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-8gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"nvidia-8gb-files-ollama","machineId":"nvidia-8gb","machine":{"id":"nvidia-8gb","label":"NVIDIA PC, 8 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":8,"minimumSystemRamGB":16},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/nvidia-8gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-8gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"nvidia-12gb-write-lm-studio","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-12gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 12 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 12 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-12gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-12gb-write-ollama","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-12gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"systemRamGB":16,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 12 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 12 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-12gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-12gb-image-lm-studio","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-12gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 12 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 12 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-12gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"nvidia-12gb-image-ollama","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-12gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"systemRamGB":16,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 12 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 12 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-12gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"nvidia-12gb-files-lm-studio","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-12gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"systemRamGB":16,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 16 GB system RAM is an explicit Bright review criterion, separately from 12 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 12 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-12gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"nvidia-12gb-files-ollama","machineId":"nvidia-12gb","machine":{"id":"nvidia-12gb","label":"NVIDIA PC, 12 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":12,"minimumSystemRamGB":16},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/nvidia-12gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-12gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"nvidia-16gb-write-lm-studio","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-16gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 16 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-16gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-16gb-write-ollama","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-16gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 16 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-16gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-16gb-image-lm-studio","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-16gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 16 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-16gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"nvidia-16gb-image-ollama","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-16gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 16 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-16gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"nvidia-16gb-files-lm-studio","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-16gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 16 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-16gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 16 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"nvidia-16gb-files-ollama","machineId":"nvidia-16gb","machine":{"id":"nvidia-16gb","label":"NVIDIA PC, 16 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":16,"minimumSystemRamGB":32},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/nvidia-16gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-16gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]},{"id":"nvidia-24gb-write-lm-studio","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"write","jobLabel":"Write and summarize","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-24gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/write","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 24 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-24gb/write","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-24gb-write-ollama","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"write","jobLabel":"Write and summarize","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-24gb/write","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/write?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B is the shared writing starting option because its exact local package and task documentation support one short exercise across these reviewed profiles. A larger download is not a reason to promote another model.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]}],"copyText":"Write and summarize: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 24 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-24gb/write?runtime=ollama","exercise":"Fictional practice note: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. Rewrite this as a friendly invitation of at most 80 words. Then list the four practical facts you preserved. Do not invent an address, contact detail or activity.\nCompare the day and time, location, rain plan and RSVP deadline with the input.\nCheck the invitation length and remove invented details before using it.\nStop or revise the prompt if facts disappear, extra facts appear or output repeats.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context."]},{"id":"nvidia-24gb-image-lm-studio","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"image","jobLabel":"Ask about an image","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-24gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/image","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B has a documented image-input path and an exact matching GGUF projector. Use this consistent starting pair and check visible details before considering larger alternatives.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_download","lm_import","lm_basics","lm_settings","lm_template","lm_vision_workflow","lm_offline","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://lmstudio.ai/blog/lmstudio-v0.3.0","https://lmstudio.ai/docs/app/offline","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Verify the matching projector and attach one picture","text":"Verify that the app associates the exact matching projector listed in this answer with the selected model. Automatic association is not guaranteed here. Attach your own non-sensitive picture through the available image attachment control and confirm it appears before sending. Current control labels and this exact pairing are untested. Stop if images are unavailable, the projector is missing or the app reports unsupported input.","sourceIds":["lm_vision_workflow","lm_qwen"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 24 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-24gb/image","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the exact matching projector. Image replies may invent details. This exercise does not establish medical or safety interpretation."]},{"id":"nvidia-24gb-image-ollama","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"image","jobLabel":"Ask about an image","runtime":"ollama","status":"starting-option","canonicalPath":"/run/nvidia-24gb/image","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/image?runtime=ollama","indexable":false,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_ollama_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},"editorialReason":"Qwen3.5 4B has a separately documented local Ollama vision package. Use the exact recorded tag in a current compatible Ollama runtime and check visible details. This mutable package is not the pinned LM Studio GGUF pair.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Windows or Linux x64 with supported NVIDIA GPU and current compatible Ollama. Exact OS minimum for this artifact is not established.","processor":"x64 CPU supported by the current Ollama distribution; confirm actual hardware support","backend":"cuda","driver":"NVIDIA compute capability 5.0+ with driver 550+; capabilities 5.0 through 6.2 require driver 570+. Check the actual GPU against official support.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","qwen35_4b_q4_ollama","ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release","lm_vision_workflow"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_ollama","title":"Qwen3.5 4B Q4_K_M: separately documented Ollama package","url":"https://ollama.com/library/qwen3.5:4b-q4_K_M","checkedAt":"2026-10-12"},{"id":"ollama_gpu","title":"Ollama GPU and driver support","url":"https://docs.ollama.com/gpu","checkedAt":"2026-10-12"},{"id":"ollama_macos","title":"Ollama on macOS","url":"https://docs.ollama.com/macos","checkedAt":"2026-10-12"},{"id":"ollama_windows","title":"Ollama on Windows","url":"https://docs.ollama.com/windows","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"ollama_qwen_release","title":"Ollama 0.17.5: small Qwen3.5 and GGUF support","url":"https://github.com/ollama/ollama/releases/tag/v0.17.5","checkedAt":"2026-10-12"},{"id":"lm_vision_workflow","title":"LM Studio: documented image attachment workflow","url":"https://lmstudio.ai/blog/lmstudio-v0.3.0","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://ollama.com/library/qwen3.5:4b-q4_K_M","https://docs.ollama.com/gpu","https://docs.ollama.com/macos","https://docs.ollama.com/windows","https://docs.ollama.com/context-length","https://github.com/ollama/ollama/releases/tag/v0.17.5","https://lmstudio.ai/blog/lmstudio-v0.3.0"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check the separate Ollama conditions","text":"Use only the exact local Ollama identifier shown in this answer and its reviewed backend conditions. Ollama packages are separate from LM Studio files. Do not substitute a cloud tag, draft package or different quantization.","sourceIds":["ollama_gpu","ollama_macos","ollama_windows"]},{"title":"Start with a short local task","text":"Use the selected local model with 4096 tokens of context and one request at a time. This is an editorial starting setting, not measured memory fit. Follow the linked Ollama package documentation and supported runtime controls using the exact identifier above. Bright has not verified the current graphical controls or image attachment workflow.","sourceIds":["ollama_context"]},{"title":"Check the result yourself","text":"Use the fictional exercise below and compare the answer with the input. Stop if loading fails, the system becomes unresponsive or the answer invents facts. Successful preparation is not proof of successful execution.","sourceIds":["ollama_context"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["ollama_context"]},{"title":"Image answers need verification","text":"An exact model and matching projector are required. Current app association and image controls are untested. Compare descriptions with the image; do not use this starter task for medical, identity or safety decisions.","sourceIds":["lm_vision_workflow"]}],"copyText":"Ask about an image: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 24 GB dedicated VRAM\nRuntime: qwen3.5:4b-q4_K_M\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-24gb/image?runtime=ollama","exercise":"Describe the two visible shapes and their colors. Separate what you can see from anything uncertain. Do not guess a purpose or add objects.\nMake your own picture of a red square beside a blue circle and attach it before asking.\nCompare every assertion with the actual picture. Confidence does not establish accuracy.\nDo not use this exercise for medical interpretation, identity or safety decisions.","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Use the separately documented Ollama vision package. No user-managed GGUF projector is specified for this path. Image replies may invent details; this exercise does not establish medical or safety interpretation."]},{"id":"nvidia-24gb-files-lm-studio","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"files","jobLabel":"Chat with a document","runtime":"lm-studio","status":"starting-option","canonicalPath":"/run/nvidia-24gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/files","indexable":true,"candidateId":"qwen35_4b_q4","candidate":{"id":"qwen35_4b_q4","name":"Qwen3.5 4B Q4_K_M","makerCheckpoint":"Qwen/Qwen3.5-4B","makerModelCardUrl":"https://huggingface.co/Qwen/Qwen3.5-4B","license":"Apache-2.0","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"Small current multimodal option with a precise community GGUF pair and explicit local Ollama tag.","runtimeEvidence":[{"id":"qwen35_4b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"qwen35_4b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"qwen3.5:4b-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"3.3GB","listedManifestShortId":"d8b0f5e9760c","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"Official release adds small Qwen3.5 sizes including 4B, fixes GGUF import and mixed CPU/GPU crashes. This establishes family/format support, not compatibility of later tag digests with this old version.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"taskSourceIds":["qwen35_4b_q4_maker","qwen35_4b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_qwen_release"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidenceId":"qwen35_4b_q4_lm-studio_cuda","runtimeEvidence":{"id":"qwen35_4b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/Qwen3.5-4B-GGUF","publisher":"lmstudio-community","revision":"f9f88ac3e234be915e23811a6d28ea287bdb927e","files":[{"name":"Qwen3.5-4B-Q4_K_M.gguf","role":"language-model","bytes":2707513696,"sha256":"25082a7dd3776cc3c741c6347d3bd04523f05796607b3fbc32fa3a25dfa1418c","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/Qwen3.5-4B-Q4_K_M.gguf"},{"name":"mmproj-Qwen3.5-4B-BF16.gguf","role":"projector","bytes":675568768,"sha256":"ae08d9d7eceb8f2d0672d61b5e6aa78b611f2942b55ed71d21414980cc454b91","listingUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/blob/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/Qwen3.5-4B-GGUF/resolve/f9f88ac3e234be915e23811a6d28ea287bdb927e/mmproj-Qwen3.5-4B-BF16.gguf"}],"transferBytes":3383082464,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.6 documents a Qwen3.5 document-chat template fix, and a maintainer confirms model support in llama.cpp engine 2.5.1. Individual loading problems were also reported. These older examples do not establish the oldest compatible version for this exact file pair.","licenseSourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact"],"taskSourceIds":["qwen35_4b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"editorialReason":"Qwen3.5 4B supplies the generative chat model for the documented LM Studio document workflow. The app handles text extraction and retrieval; an embedding-only model is not the chat choice.","conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"evidenceState":"documentation-checked","hardwareAssessment":"estimated","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":["apache","qwen35_4b_q4_maker","qwen35_4b_q4_artifact","lm_system","lm_runtime","lm_current_runtime","lm_qwen","lm_qwen_rag_fix","lm_qwen_engine_statement","ollama_context","lm_docs","lm_offline","lm_download","lm_import","lm_basics","lm_settings","lm_template","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"sources":[{"id":"apache","title":"Apache License 2.0 terms","url":"https://www.apache.org/licenses/LICENSE-2.0","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_maker","title":"Qwen3.5 4B Q4_K_M: maker checkpoint and license","url":"https://huggingface.co/Qwen/Qwen3.5-4B","checkedAt":"2026-10-12"},{"id":"qwen35_4b_q4_artifact","title":"Qwen3.5 4B Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_system","title":"LM Studio system requirements","url":"https://lmstudio.ai/docs/app/system-requirements","checkedAt":"2026-10-12"},{"id":"lm_runtime","title":"LM Studio app and runtime documentation","url":"https://lmstudio.ai/docs/app","checkedAt":"2026-10-12"},{"id":"lm_current_runtime","title":"LM Studio current app and engine release notes","url":"https://lmstudio.ai/changelog/lmstudio","checkedAt":"2026-10-12"},{"id":"lm_qwen","title":"LM Studio Qwen3.5 model information","url":"https://lmstudio.ai/models/qwen3.5","checkedAt":"2026-10-12"},{"id":"lm_qwen_rag_fix","title":"LM Studio 0.4.6: Qwen3.5 document-chat template fix","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","checkedAt":"2026-10-12"},{"id":"lm_qwen_engine_statement","title":"LM Studio maintainer: Qwen3.5 engine support and reported loading issues","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","checkedAt":"2026-10-12"},{"id":"ollama_context","title":"Ollama context length and memory use","url":"https://docs.ollama.com/context-length","checkedAt":"2026-10-12"},{"id":"lm_docs","title":"LM Studio: chat with documents","url":"https://lmstudio.ai/docs/app/basics/rag","checkedAt":"2026-10-12"},{"id":"lm_offline","title":"LM Studio offline operation","url":"https://lmstudio.ai/docs/app/offline","checkedAt":"2026-10-12"},{"id":"lm_download","title":"LM Studio: downloading the selected model","url":"https://lmstudio.ai/docs/app/basics/download-model","checkedAt":"2026-10-12"},{"id":"lm_import","title":"LM Studio: importing local GGUF files","url":"https://lmstudio.ai/docs/app/advanced/import-model","checkedAt":"2026-10-12"},{"id":"lm_basics","title":"LM Studio: desktop setup","url":"https://lmstudio.ai/docs/app/basics","checkedAt":"2026-10-12"},{"id":"lm_settings","title":"LM Studio per-model settings","url":"https://lmstudio.ai/docs/app/advanced/per-model","checkedAt":"2026-10-12"},{"id":"lm_template","title":"LM Studio: prompt templates","url":"https://lmstudio.ai/docs/app/advanced/prompt-template","checkedAt":"2026-10-12"},{"id":"gemma_license","title":"Google Gemma Apache 2.0 licensing","url":"https://ai.google.dev/gemma/apache_2","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_maker","title":"Gemma 4 12B IT Q4_K_M: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_12b_q4_artifact","title":"Gemma 4 12B IT Q4_K_M: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","checkedAt":"2026-10-12"},{"id":"lm_gemma","title":"LM Studio Gemma 4 model information","url":"https://lmstudio.ai/models/gemma-4","checkedAt":"2026-10-12"},{"id":"lm_gemma_template","title":"LM Studio 0.4.11: Gemma 4 template support","url":"https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","checkedAt":"2026-10-12"},{"id":"lm_gemma12_engine_fix","title":"LM Studio maintainer: Gemma 4 12B engine loading fix","url":"https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_maker","title":"Gemma 4 26B-A4B IT QAT Q4_0: maker checkpoint and license","url":"https://ai.google.dev/gemma/docs/core/model_card_4","checkedAt":"2026-10-12"},{"id":"gemma4_26b_qat_artifact","title":"Gemma 4 26B-A4B IT QAT Q4_0: exact GGUF publisher metadata","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true","checkedAt":"2026-10-12"}],"sourceUrls":["https://www.apache.org/licenses/LICENSE-2.0","https://huggingface.co/Qwen/Qwen3.5-4B","https://huggingface.co/api/models/lmstudio-community/Qwen3.5-4B-GGUF/revision/f9f88ac3e234be915e23811a6d28ea287bdb927e?blobs=true","https://lmstudio.ai/docs/app/system-requirements","https://lmstudio.ai/docs/app","https://lmstudio.ai/changelog/lmstudio","https://lmstudio.ai/models/qwen3.5","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.6","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/1606","https://docs.ollama.com/context-length","https://lmstudio.ai/docs/app/basics/rag","https://lmstudio.ai/docs/app/offline","https://lmstudio.ai/docs/app/basics/download-model","https://lmstudio.ai/docs/app/advanced/import-model","https://lmstudio.ai/docs/app/basics","https://lmstudio.ai/docs/app/advanced/per-model","https://lmstudio.ai/docs/app/advanced/prompt-template","https://ai.google.dev/gemma/apache_2","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/lmstudio-community/gemma-4-12B-it-GGUF/revision/40b870babe39398ce917bbf4b4ff9b5a1f12710e?blobs=true","https://lmstudio.ai/models/gemma-4","https://lmstudio.ai/changelog/lmstudio/lmstudio-v0.4.11","https://github.com/lmstudio-ai/lmstudio-bug-tracker/issues/2004","https://ai.google.dev/gemma/docs/core/model_card_4","https://huggingface.co/api/models/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/revision/d1c082be9cf3c8a514acf63b8761f4b41935842e?blobs=true"],"reviewedAt":"2026-10-12","changedAt":"2026-10-12","publishedAt":"2026-10-12","revision":"2026-10-12-r1","history":[{"revision":"2026-10-12-r1","date":"2026-10-12","reason":"Initial documentation-led editorial starting selections. No hands-on hardware tests."}],"setup":[{"title":"Check your machine conditions","text":"Read the separate OS, processor, backend, system RAM and dedicated VRAM conditions shown in this answer. Installed memory is not available memory. Apple Silicon uses unified memory; never add system RAM and VRAM to establish compatibility.","sourceIds":["lm_system"]},{"title":"Install and update LM Studio","text":"Install the current stable desktop app and update its compatible llama.cpp runtime. The source-check example is LM Studio 0.4.26 Build 4; Build 2 documents runtime 2.50.0. These are documented examples, not exact-artifact success guarantees. Use GGUF with the matching Metal or CUDA backend in your conditions; this answer does not provide an MLX path.","sourceIds":["lm_runtime","lm_current_runtime"]},{"title":"Select the exact reviewed files","text":"In Discover, search the exact publisher and repository shown above. Match every selected filename and reviewed revision, including any required projector. Do not choose a similarly named curated package. If the listing cannot establish that revision, use the linked pinned files and documented import route, or stop for verification.","sourceIds":["lm_download","lm_import"]},{"title":"Load one model with a short context","text":"Open Chat and its model loader, then select the downloaded model. Set context to 4096 tokens and use one request at a time. These are Bright editorial starting settings, not maker maxima or measured fit. My Models' settings or the pre-load configuration can set context. Keep the metadata chat template; current GGUF template settings are under Load Parameters > Advanced, but the exact current graphical labels have not been inspected.","sourceIds":["lm_basics","lm_settings","lm_template","lm_current_runtime"]},{"title":"Keep Qwen3.5 reasoning settings model-specific","text":"Do not copy Qwen3 /think or /nothink switches. Qwen3.5 documents the enable_thinking chat-template setting. Keep the supplied metadata template and use an app-supported reasoning control only if available and understood. The exact desktop control and behavior are untested.","sourceIds":["qwen35_4b_q4_maker","lm_template"]},{"title":"Attach a short text-based document","text":"Create the fictional picnic-practice.txt below and drag it into a fresh LM Studio chat with the generative model. The documented attachment formats are .txt, .pdf and .docx. Short extracted text can enter context in full; longer documents use retrieval that can miss relevant passages. An embedding-only model is not the chat pick.","sourceIds":["lm_docs"]},{"title":"Try the fictional task and verify the answer","text":"Use the exercise below. Compare every claimed fact or quotation with the original input. Stop if loading or extraction fails, memory pressure or swap becomes high, the system becomes unresponsive, or output repeats. Shorten input or choose a smaller reviewed option rather than forcing a fit.","sourceIds":["lm_system","lm_docs"]},{"title":"Keep the local-only conditions visible","text":"After the model files and runtime are installed, local model and document chat are documented to work offline. Search, downloads and updates use the network. Select the installed local model and avoid remote providers, LM Link, cloud services, external tools and plugins for this path. Their behavior is outside this offline claim.","sourceIds":["lm_offline"]},{"title":"Estimated and untested","text":"Download bytes do not establish runtime memory or compatibility. Peak memory, speed and exact hardware execution remain untested. Historical application and runtime support examples are not blanket guarantees for these artifacts.","sourceIds":["lm_system"]},{"title":"Retrieval has limits","text":"Start with a short text-based document. Scanned PDFs may need separate OCR. Tables, figures, handwriting and full page coverage are unverified. Vision capability does not prove that the attachment extractor reads scanned pages. Verify supporting passages in the actual file; no page-count promise is made.","sourceIds":["lm_docs"]}],"copyText":"Chat with a document: Qwen3.5 4B Q4_K_M\nNVIDIA PC, 24 GB dedicated VRAM\nRuntime: lmstudio-community/Qwen3.5-4B-GGUF\nStarter context: 4096 tokens; one chat\nDocumentation checked 2026-10-12; estimated and untested\nhttps://brightaifuture.com/run/nvidia-24gb/files","exercise":"In picnic-practice.txt, what happens if it rains, and who provides lunch? Quote the exact supporting sentences for each answer. If the file does not say something, say it is not stated.\nCreate picnic-practice.txt: The Maple Street club's picnic is Saturday at noon in Riverside Park. Bring your own lunch. If it rains, meet in the community hall. RSVP to the organizer by Thursday. The club will not provide food.\nCheck every quoted sentence against the file, then ask for the absent organizer phone number and check that the answer says it is not stated.\nStop if extraction fails or supporting quotations are invented. A quotation is something to verify.","alternatives":[{"candidate":{"id":"gemma4_12b_q4","name":"Gemma 4 12B IT Q4_K_M","makerCheckpoint":"google/gemma-4-12B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger dense model for the reviewed higher-memory profiles. It is not offered for the 16 GB Mac or 12 GB NVIDIA profiles; download size alone does not establish compatibility.","runtimeEvidence":[{"id":"gemma4_12b_q4_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_12b_q4_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:12b-it-q4_K_M","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"8.0GB","listedManifestShortId":"6114515d63c1","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"taskSourceIds":["gemma4_12b_q4_maker","gemma4_12b_q4_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"needs-verification","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag. This package includes a 423M Q8_0 draft model and BF16 projector; LM Studio pair has no added draft model.","The selected 12B Ollama package includes a draft model. Its compatible CUDA runtime version has not been verified, so it is not offered as a starting option or alternative.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_12b_q4_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"lmstudio-community/gemma-4-12B-it-GGUF","publisher":"lmstudio-community","revision":"40b870babe39398ce917bbf4b4ff9b5a1f12710e","files":[{"name":"gemma-4-12B-it-Q4_K_M.gguf","role":"language-model","bytes":7381382944,"sha256":"95d83ba36642b1f385fb906b5962a71763361be3bac930a709945f72d97473f8","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/gemma-4-12B-it-Q4_K_M.gguf"},{"name":"mmproj-gemma-4-12B-it-BF16.gguf","role":"projector","bytes":175115296,"sha256":"7fd884a5f7d9ee60f6b88e5b51287151e547454b2b60a4a6270bb5434310ad0e","listingUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/blob/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf","downloadUrl":"https://huggingface.co/lmstudio-community/gemma-4-12B-it-GGUF/resolve/40b870babe39398ce917bbf4b4ff9b5a1f12710e/mmproj-gemma-4-12B-it-BF16.gguf"}],"transferBytes":7556498240,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. A maintainer reports that llama.cpp engine 2.20.1 fixes a Gemma 4 12B loading problem observed with engine 2.20.0. The exact oldest compatible version for this non-draft file pair is not established.","licenseSourceIds":["apache","gemma_license","gemma4_12b_q4_maker","gemma4_12b_q4_artifact"],"taskSourceIds":["gemma4_12b_q4_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented dense Gemma alternative on the reviewed higher-memory scope; LM Studio uses a non-draft file pair. No better-quality or speed claim."},{"candidate":{"id":"gemma4_26b_qat","name":"Gemma 4 26B-A4B IT QAT Q4_0","makerCheckpoint":"google/gemma-4-26B-A4B-it","makerModelCardUrl":"https://ai.google.dev/gemma/docs/core/model_card_4","license":"Apache-2.0","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"jobs":["write","files","image"],"starterContextTokens":4096,"contextProvenance":"Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","selectionReason":"An optional larger mixture-of-experts model. Active parameter count does not establish total memory use. This QAT file differs from the original full-precision checkpoint, whose benchmark results do not establish this file’s performance.","runtimeEvidence":[{"id":"gemma4_26b_qat_lm-studio_metal","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"metal","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_metal","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"metal","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]},{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},{"id":"gemma4_26b_qat_ollama_cuda","runtime":"ollama","exactIdentifier":"gemma4:26b-a4b-it-qat","publisher":"Ollama registry","revision":null,"files":[],"transferBytes":null,"listedTransferSize":"16GB","listedManifestShortId":"2dd70431afed","immutablePin":false,"backend":"cuda","appVersion":null,"engineVersion":null,"versionExplanation":"0.30.5 explicitly fixes gemma4:12b floating-point crash on x86/CUDA/Linux/Windows. 0.30.6 explicitly names gemma4:26b-a4b-it-qat. Later exact 12B package now carries a draft model; these historical releases alone do not establish its minimum.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"taskSourceIds":["gemma4_26b_qat_maker","gemma4_26b_qat_ollama"],"compatibilitySourceIds":["ollama_gpu","ollama_macos","ollama_windows","ollama_context","ollama_gemma12_fix","ollama_gemma_qat"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Tags can change. Recorded Ollama package is a separate build; not asserted identical to LM Studio GGUF. Never substitute a -cloud tag.","Short manifest IDs and mutable tags are not immutable artifact pins. LM Studio file hashes and sizes do not establish this package’s bytes or memory."]}],"checkedAt":"2026-10-12"},"runtimeEvidence":{"id":"gemma4_26b_qat_lm-studio_cuda","runtime":"lm-studio","exactIdentifier":"google/gemma-4-26B-A4B-it-qat-q4_0-gguf","publisher":"google","revision":"d1c082be9cf3c8a514acf63b8761f4b41935842e","files":[{"name":"gemma-4-26B_q4_0-it.gguf","role":"language-model","bytes":14439363584,"sha256":"3eca3b8f6d7baf218a7dd6bba5fb59a56ee25fe2d567b6f5f589b4f697eca51d","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B_q4_0-it.gguf"},{"name":"gemma-4-26B-it-mmproj.gguf","role":"projector","bytes":1194828160,"sha256":"a359953a076b877db30c31dbbb4c6d93b4a6e017ee5db5784247e4d4c0dd4f3b","listingUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/blob/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf","downloadUrl":"https://huggingface.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf/resolve/d1c082be9cf3c8a514acf63b8761f4b41935842e/gemma-4-26B-it-mmproj.gguf"}],"transferBytes":15634191744,"listedTransferSize":null,"listedManifestShortId":null,"immutablePin":true,"backend":"cuda","appVersion":"0.4.26 Build 2 or later","engineVersion":"llama.cpp 2.50.0 (upstream b11337)","versionExplanation":"The app and engine versions are documented. Support for these exact file hashes is inferred from documented model support and the later engine, not directly tested. LM Studio 0.4.11 documents updated Gemma 4 templates. The recorded Gemma 4 12B loading fix in engine 2.20.1 concerns a different model and does not prove support for this 26B QAT file pair. Its exact oldest compatible version remains unknown.","licenseSourceIds":["apache","gemma_license","gemma4_26b_qat_maker","gemma4_26b_qat_artifact"],"taskSourceIds":["gemma4_26b_qat_maker"],"compatibilitySourceIds":["lm_system","lm_runtime","lm_current_runtime","lm_gemma","lm_gemma_template","lm_gemma12_engine_fix"],"evidenceState":"documentation-checked","hardwareTestReceipt":null,"eligibility":"reviewed","limitations":["Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option."]},"conditions":{"platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"systemRamGB":32,"operatingSystem":"Reviewed Windows x64 app path, or Ubuntu 20.04+ x64 AppImage. Vendor says Ubuntu versions newer than 22 are not well tested. Exact Windows minimum for this artifact is not established.","processor":"x64 CPU with AVX2; integrated-only graphics excluded","backend":"cuda","driver":"NVIDIA GPU and driver compatible with the selected LM Studio CUDA runtime. Exact minimum driver for these artifacts is not established; verify the current engine and actual GPU before loading.","starterContextTokens":4096,"concurrency":1,"qualification":["This is the reviewed setup scope, not hardware detected from your browser and not a promise that the model loads or performs well.","Confirm installed and currently available memory separately; use one chat at a time and start with the stated context.","At least 32 GB system RAM is an explicit Bright review criterion, separately from 24 GB dedicated VRAM. It does not establish available memory or fit.","Record app and runtime/backend versions. Use current stable compatible runtime; historical version examples are not oldest-supported-version guarantees."]},"reason":"Optional separately documented QAT MoE alternative on the reviewed higher-memory scope. Active parameters do not represent total memory; no benchmark transfer or fit claim."}],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum.","Exact oldest compatible version of these pinned artifacts is not established. Current version mapping and historical family support are documentary evidence, not a hardware test.","Discover may select newer files. Compare publisher, filenames and recorded revision before treating a changed package as this reviewed option.","Start with 4096 tokens for this exercise. The model supports longer contexts, but longer conversations use more memory. This starting setting is not a tested memory limit or the model’s maximum context.","Start with a short text-based document. Scans, tables, figures and handwriting may not extract correctly; retrieval can miss passages. Check every supporting passage in the original. Offline operation excludes enabled remote tools, providers, plugins or cloud tags."]},{"id":"nvidia-24gb-files-ollama","machineId":"nvidia-24gb","machine":{"id":"nvidia-24gb","label":"NVIDIA PC, 24 GB dedicated VRAM","platform":"nvidia","unifiedMemoryGB":null,"dedicatedVramGB":24,"minimumSystemRamGB":32},"job":"files","jobLabel":"Chat with a document","runtime":"ollama","status":"unsupported","canonicalPath":"/run/nvidia-24gb/files","canonicalUrl":"https://brightaifuture.com/run/nvidia-24gb/files?runtime=ollama","indexable":false,"candidateId":null,"candidate":null,"runtimeEvidenceId":null,"runtimeEvidence":null,"editorialReason":"Direct Ollama chat is not the verified document-attachment and retrieval workflow. Choose LM Studio for this task.","conditions":null,"evidenceState":null,"hardwareAssessment":"not-established","runtimeMemoryBytes":null,"speedTokensPerSecond":null,"sourceIds":[],"sources":[],"sourceUrls":[],"reviewedAt":null,"changedAt":null,"publishedAt":null,"revision":"2026-10-12-r1","history":[],"setup":[],"copyText":"","exercise":"","alternatives":[],"limitations":["Documentation-checked and estimated. Bright has not loaded these models or tested these computer configurations. Runtime memory, speed and output quality are unknown.","Download bytes are transfer size, not runtime memory. Available memory differs from installed memory. Context, app/runtime, projector, other applications and load settings affect use.","System RAM and dedicated VRAM are separate resources. Do not add them or infer compatibility from their sum."]}]}