Mistral Releases Weights for Trillion-Parameter Large 4 Model, Targeting Sovereign Cyber Defense

French AI company Mistral has released the weights for Mistral Large 4, its trillion-parameter open source model nicknamed “Le Chonk,” allowing businesses and governments to download and deploy it themselves. The model, first introduced earlier this month, is available on Hugging Face and through Mistral’s API.

The release sharpens the contest between open-weight developers and closed-model providers such as OpenAI and Anthropic. CEO Arthur Mensch has argued that AI should empower rather than dominate, and Mistral says ML4 achieves state-of-the-art performance using a fraction of the compute and capital of closed labs.

Mistral said the model matches Chinese open-weight rival GLM-5.3, particularly on cyber tasks, while using 47% fewer tokens per task, and that it leads all open models on Artificial Analysis’ Coding Agent Index for agentic coding. The company positioned ML4 as a sovereign tool for detecting and responding to threats across code and IT infrastructure.

Pierre Stock, Vice President of Science, said open weights give organizations access to frontier AI for cyber defense that cannot be cut off, and allow on-premises deployment. He said this capability should reach everyone quickly, as malicious actors may use their own AI to attack. Addressing rogue agent incidents, Stock cautioned against fearmongering, describing AI security as a hard but solvable problem.