discovery · Source published 2024-11-15
Assembling biological model workflows
BioNeMo brings biological foundation models and training components into one framework for protein, molecular, and drug-research experiments.
Find your way in.
106 items for “open models”: records, experiences, guides and model families. Source reports remain distinct from discoveries.
discovery · Source published 2024-12-03
Adapting one Earth model to many maps
Prithvi-EO-2.0 is a geospatial foundation model that can be adapted to Earth-observation tasks such as flood mapping, crop classification, and land-use analysis.
discovery · Source published 2024-02-29
Adapting a code model to a private codebase
StarCoder2 offers several downloadable code-model sizes plus fine-tuning and evaluation examples for local completion and generation experiments.
discovery · Source published 2025-03-13
A model whose full recipe is visible.
Ai2 released OLMo 2 32B, a language model accompanied by publicly available data, code, weights, training details, and a reproducible training recipe.
discovery · Source published 2025-03-18
Reasoning models with more of the toolkit in view.
NVIDIA announced the Llama Nemotron Nano and Super reasoning models, alongside plans to make the post-training tools, datasets and optimization techniques openly available.
discovery · Source published 2025-04-29
Choosing when a model thinks longer.
The Qwen team released and described Qwen3, a family of dense and mixture-of-experts language models with selectable thinking and non-thinking modes.
discovery · Source published 2026-09-14
The forecast competition leaves its data open.
ECMWF released JJA 2026 forecast files and regional scores for an open framework comparing AI and hybrid subseasonal weather systems.
discovery · Source published 2026-05-07
An Earth model runs above the Earth.
A compressed NASA–IBM Prithvi geospatial model ran on the Kanyini satellite and the ISS IMAGIN-e payload in flood and cloud detection demonstrations.
discovery · Source published 2026-10-19
GoTo’s downloadable AI model makes room for Indonesia’s languages
The Gemma-based model lists Indonesian, English and four regional languages. Its public weights offer builders more choice, while language quality, computing needs and license terms still matter.
guide · Source published date unknown
How to run an AI model on your own computer
Built from LM Studio’s and the model makers’ documentation, checked 11 October 2026. Bright has not tested this on its own hardware.
discovery · Source published 2026-09-23
A Swedish speech model brings fast transcription closer to home.
Klang’s released speech model offers a practical choice about where Swedish audio is processed. Its benchmark gains are promising, with important differences between speed and accuracy.
discovery · Source published 2026-10-27
Liquid AI releases small models for local text, vision and voice tasks
LFM3 brings downloadable models to everyday devices, with a focus on specialized applications. Hardware requirements, software integration and a custom commercial license still shape what developers can build.
discovery · Source published 2026-10-10
Reflection AI’s Beam could give developers a powerful new open AI option
Reflection AI’s Beam preview could expand open-weight AI. What its planned release and reported NVIDIA talks could mean for developers and researchers.
discovery · Source published 2026-10-20
DeepSeek opens the engine behind its AI. Here’s what DeepInfer actually changes
DeepSeek has opened DeepInfer, the engine behind V4.1-Flash. A plain-language guide to the release, its benchmarks, hardware needs and limits.
model-family · Source published date unknown
Qwen3.8: license, size and what is open
A downloadable 27B vision-language member of the Qwen3.8 generation, selected here because its Apache-2.0 terms are materially simpler than the custom license on the largest Qwen3.8 checkpoint.
model-family · Source published 2026-04-02
Gemma 4: license, size and what is open
Google’s Gemma 4 generation spans edge and workstation sizes; this atlas record uses the 26B mixture-of-experts checkpoint as its reference point.
model-family · Source published date unknown
Llama 4: license, size and what is open
Meta’s first natively multimodal Llama generation and first Llama mixture-of-experts release.
model-family · Source published date unknown
GLM-5.3 Flash: license, size and what is open
A natively multimodal sparse GLM checkpoint released under MIT terms for long-context agent and coding work.
discovery · Source published 2022-08-22
Adapting an image model to a visual practice
Stable Diffusion's downloadable latent-diffusion weights and adaptation code allowed artists and developers to build local image generation, inpainting, fine-tuning, and LoRA workflows.
discovery · Source published 2025-04-28
A security model for a SOC's own evidence
Foundation-Sec-8B is a cybersecurity-focused Llama 3.1 derivative that organizations can download and adapt for security operations work involving their own alerts, cases, and threat knowledge.