Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity
techcrunch ·
French AI company Mistral has released the open-source weights for its trillion-parameter model, Mistral Large 4 (ML4), also known as "Le Chonk." This release allows businesses and governments to download and deploy the system independently, emphasizing data sovereignty in cybersecurity. The model is positioned as a cost-effective alternative to closed providers, using 47 percent fewer tokens than competitor GLM-5.3 while maintaining state-of-the-art performance. Mistral claims ML4 leads all open models in agentic coding tasks on the Artificial Analysis Coding Agent Index. By making the weights available on Hugging Face and via API, the company aims to empower organizations with sovereign cyber defense capabilities. Executives argue that open-weight access ensures control over AI tools used to detect and respond to threats without reliance on external vendors. Despite concerns about rogue AI agents, Mistral maintains that security risks are solvable and urges widespread adoption to strengthen the ecosystem against malicious actors.
techcrunch