Mistral Large 4: a trillion parameters, with the weights due this month

The French company Mistral has opened a public preview of Mistral Large 4, a natively multimodal model with 1 trillion parameters, of which 49 billion run for any one word, and a 1 million token context window. It was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European datacentres, on data spanning more than 160 languages. Mistral says it will publish the open weights by the end of October 2026. Its most striking claim is in security: on a test that asks a model to reproduce a real software vulnerability and then patch it, Mistral Large 4 scores 82%, the highest of any model, while Claude Opus 5.5 and GPT-6 Astra score near zero because they refuse the task.

Mistral AI: Mistral Large 4 · 6 October 2026

We write news backed by research from different sources. Quick news is a short summary that links to the original announcement. How we report