OLMo 3.1
OLMo 3.1
The only model here where you can read the training data as well as the weights.
Updated
The read
Ai2 publishes everything: weights, the full training corpus, the code, and the intermediate checkpoints from along the run. No other model in this index comes close to that, and for research, auditing, or any deployment that has to answer "what was it trained on", nothing else is a real option. On raw capability it does not compete with the models it sits beside, and it is not pretending to.
Where it shines
- Genuinely reproducible: data and checkpoints published, not just weights
- Apache 2.0 throughout, with no downstream conditions
- The only credible answer when provenance is a hard requirement
Worth knowing first
- Clearly behind the field on general capability
- Short context window by current standards
Index score
72.0
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 46 of 53
- Builds
- LLMs
- Category
- Open weights
- Output
- Apache-licensed weights, data, code and checkpoints
- Pricing
- Free tier
- Best for
- Research and any deployment that must audit its training data
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Also worth a look
Full index- 05Kimi K3Kimi K3ProvisionalThe largest open-weights model anyone has shipped, and it reads a million tokens without blinking.Open weightsFree tier87.5
- 06GLM-5.2GLM-5.2ProvisionalFrontier-class output, an MIT licence, and a million tokens of context. That combination is the story.Open weightsFree tier87.1
- 12Qwen 3.6Qwen 3.6ProvisionalThe open Qwen release most people should actually run, in sizes that fit real hardware.Open weightsFree tier83.3
