Mistral Large 4 Ships With 1.05T Open Weights and 1M Context
- Source
- x.com
- Date
Mistral Large 4 joins the model lineup on Open Math OS. 1.05T open weights, 1M context, live today for approved teams. Open Math OS is a research workspace where mathematicians and AI agents work on shared projects, built only on open models so the community can inspect and run them, and with the community in control of its own data. More open frontier models in the lineup means more choice for every team.
Context
Mistral AI's November 3 post announces a new place to use an existing model, not a new model: Mistral Large 4 has joined the lineup of models on Open Math OS (x.com). The post describes Open Math OS as a research workspace where mathematicians and AI agents work on shared projects, built only on open models so the community can inspect and run them, with the community in control of its own data. It says Large 4 is "live today for approved teams." The post does not say who runs Open Math OS, how a team gets approved, or what Large 4 has done there yet. Replies and any linked material were not collected, so those questions remain open.
The numbers in the post, 1.05 trillion and a 1-million- , describe the model as Mistral released it days earlier. Mistral launched Large 4 on October 6 as a public preview through its API, calling it a 1 trillion-parameter model with 49 billion active parameters, and said the weights would follow by the end of the month (mistral.ai). Active parameters are the part of a model that runs for each token: the model routes every token through a small group of its experts (huggingface.co). On October 29, Artificial Analysis reported that Mistral had updated the model and released its weights at 1.05 trillion total and 52 billion active parameters, with a 1-million-token context window. Open weights mean anyone can download the model files and run them, which is the property Open Math OS says it requires. Artificial Analysis also notes a condition in the license: running a model-as-a-service business on it requires a separate commercial license from Mistral.
Mistral's launch post named math as one of the model's strengths, in both formal reasoning and applied mathematics, and said its own human evaluations found Large 4 reasoned "more precisely and with more structure" than GLM-5.3 (mistral.ai). That is Mistral's assessment of its own model. The Open Math OS post reports no results from the workspace itself.
- open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
- token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
- mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.
Checking sign-in…
Loading comments…





