Ai2✓

Stav’s assessment

Editorial assessment, not legal advice. Stav's risk ratings, scores, and verdicts are our own analysis of publicly available information and may be incomplete or out of date. Verify independently before making compliance or procurement decisions.

• transparency

  Ai2 releases full model weights, training data (Dolma/Dolma 2/Dolma 3), training code, and intermediate checkpoints under permissive licences (Apache 2.0/ODC-BY), going further than most labs' 'open-weights' releases.

• community

  Long-standing academic and industry partnerships (University of Washington, NVIDIA, Cirrascale) and continued frontier open-model releases reflect sustained community engagement.

• compliance

  Independent assessment found Ai2's model cards, datasheets, and repositories to be well-documented, described as providing a broad overview with links to full technical detail.

• stability

  Ai2 has continued shipping major open model releases (OLMo 3, OLMo 3.1) through late 2025 and into 2026, indicating an actively operating research programme despite leadership transitions.

Published safeguards & certifications