Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity

Mistral released the weights for Mistral Large 4, a trillion-parameter open-weight AI model nicknamed “Le Chonk,” enabling businesses and governments to download and deploy it on their own infrastructure. First introduced earlier in October, ML4 is available through Hugging Face and Mistral’s API. The numerical weights allow organizations to operate the model independently rather than relying on a hosted provider for access. Mistral is positioning ML4 for cybersecurity work, including detecting, analyzing and responding to threats across code and IT infrastructure. Pierre Stock, Mistral’s vice president of science, said organizations can deploy the model on-premises to identify, coordinate and respond to major security incidents in real time. Mistral said ML4 delivers state-of-the-art sovereign cyber-defense capabilities, while CEO Arthur Mensch said AI should be a tool for empowerment rather than dominance. Mistral claims ML4 has performance comparable to Chinese open-weight model GLM-5.3, particularly for cyber work, while using 47% fewer tokens per task. The company also claims ML4 leads other open models on agentic coding tasks in Artificial Analysis’ Coding Agent Index. Stock acknowledged risks from rogue AI agents but said the security problem is difficult and solvable.