French AI company Mistral has published the weights for Mistral Large 4, a trillion-parameter model it calls “Le Chonk.” The weights are on Hugging Face and the model is available through Mistral’s API, so businesses and governments can download and deploy it themselves.

Mistral pitches it as sovereign cyber defense: a way for enterprises and public-sector bodies to “download, deploy, and own” frontier AI to detect, analyze and respond to threats.

The performance claims are Mistral’s own, and the company published no numerical scores. It says Large 4 is comparable to GLM-5.3, a Chinese open-weight model, especially for cyber work, while using 47% fewer tokens per task, and that it outranks all other open models on Artificial Analysis’ Coding Agent Index. TechCrunch’s article does not name a licence.

Pierre Stock, Mistral’s vice president of science, argued that open weights give access “that can’t be cut off by definition.” He said the same capability should reach everyone quickly because malicious actors may use AI to attack, while cautioning against fearmongering about recent rogue AI agent incidents. Chief executive Arthur Mensch said AI should “be a tool for empowerment, not dominance.”

Sources

Image: La Défense skyline from Eiffel Tower.jpg by Gugalcrom123, CC BY 4.0, via Wikimedia Commons. Illustrative photo; it does not show the events described.