Introducing Mistral Large 4: Le chonk
Mistral has released a preview of Mistral Large 4, a model with 1 trillion total parameters and 49 billion active parameters. It was trained on Mistral's own cluster of 3,800 NVIDIA Grace Blackwell GPUs. The preview is available through Mistral's API, and the company says it will release the open weights model at the end of this month.
The model supports two reasoning levels through the Mistral API: "none" and "high." In one comparison cited, the "high" setting used 2,717 output tokens while "none" used 3,275.
On Artificial Analysis, Mistral Large 4 scores 38, placing it just behind DeepSeek 4.1 Flash, which is described as a 552B model. It is a large improvement over Mistral Large 3, released last December, which scored 9 on Artificial Analysis.
Mistral Large 4 is described as not a Fable-class model, but as a sign that Mistral has returned to roughly six months behind the frontier. The item was noted via Hacker News.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Introducing Mistral Large 4: Le chonk Mistral are back in the game. Today they're releasing a preview of Mistral Large 4, a 1 trillion parameter, 49 billion active parameter model trained on their own cluster of 3,800 NVIDIA Grace Blackwell GPUs. The preview is available via their API. They promise to release the open weights model at the "end of this month". The model only supports two reasoning levels - "none" and "high" - via the Mistral API. Here are both pelicans - the "high" one looks better, though surprisin