available
Weights
Multiple official checkpoints are released after pre-training, supervised fine-tuning, and long-context fine-tuning.
Inspect this claim’s source ↗Allen Institute for AI (Ai2) / FAMILY RECORD
Molmo 2
A family of open vision-language models for images, multiple images, and video, with explicit spatial outputs such as pointing and object tracking.
Checked release: Molmo2-4B, Molmo2-8B, and Molmo2-O-7B
Sources checked 2026-09-19
- Molmo 2 release (month documented; day not established)
- · Primary source
Ai2’s primary media record establishes December 2025, not a release day; this field remains null rather than inventing one.
The profile describes this checkpoint, not every release carrying the family name. Developer documentation is not independent reproduction.
THE MODEL / THE TERMS
Know what you are picking up.
- Parameters
- 4B, 7B, and 8B variants
- Architecture
- Vision-language models with variants specialized for visual understanding and spatial grounding.
- Modalities
- text · image · video · pointing coordinates · tracking coordinates
- Context
- Varies by checkpoint and training stage; use the versioned model card.
- Weights license
- Apache-2.0 repository and model terms; some source datasets require separate agreements
- Commercial use
- The released code and model artifacts use Apache-2.0, but dataset terms must be reviewed separately.
- Memory
- Requirements vary from the 4B to 8B variants and with video length; no single minimum is asserted.
- Known quantizations
- No canonical quantized checkpoint is asserted in this record
Specifications are developer-reported. Primary release documentation ↗ · All sources and terms
NINE DIMENSIONS / NO OPENNESS SCORE
What is actually available?
Each component has its own evidence. “Not established” means this source review did not verify availability; it is not a claim that nothing exists.
available
Architecture
The repository and technical report document model design and variant differences.
Inspect this claim’s source ↗available
Inference code
Inference and demo code is included in the official repository.
Inspect this claim’s source ↗available
Training code
Training scripts for the documented stages are public.
Inspect this claim’s source ↗available
Training recipe
The repository documents staged training and links the technical report.
Inspect this claim’s source ↗available
Data information
Training datasets and stage-specific inputs are identified.
Inspect this claim’s source ↗partial
Training data
Most dataset download scripts are supplied, while some datasets require separate agreements.
Inspect this claim’s source ↗available
Evaluation
Evaluation code and benchmark details are released with the repository.
Inspect this claim’s source ↗partial
Commercial use
Apache-2.0 covers released artifacts, while source datasets retain separate terms.
Inspect this claim’s source ↗WHY THIS RELEASE MATTERS
What to explore.
- Video understanding
- Multi-image reasoning
- Pointing
- Object tracking
- Open training scripts
WHAT THE RECORD DOES NOT ESTABLISH
Keep the limits.
- Some training datasets need separate agreements
- Spatial outputs still require task-specific evaluation before deployment
This record does not attach an independent reproduction of Molmo 2 results; it highlights that the released artifacts support such work.
FOLLOW THE RECORD
Sources, not superlatives.
- Ai2 media center — Molmo 2 release month ↗ (institution)
- Molmo 2 repository ↗ (repository)
- Molmo 2 model and training documentation ↗ (report)
Source check: 2026-09-19. AI-assisted editorial review, with no named external expert or independent hardware validation claimed. Licenses and model cards can change. Review the linked terms for the specific artifact and intended use.
Understand open source vs open weights ↗