Mistral AI has officially released its Large 4 model to the public. Previously accessible only during a limited public preview earlier in October, the improved version is now open weight for all users.

The company describes Large 4, also known by the codename Le Chonk, as the leading open-weight model for coding tasks. According to Mistral, the release offers significant advancements in cybersecurity-related work, specifically in generating patches and reproducing vulnerabilities.

Mistral notes that the model demonstrates better safety protocols than other open models in this category. It is designed to be more likely to refuse malicious cyber requests while still supporting legitimate security research and development activities.

The announcement was published on October 29, 2026, marking the transition from preview access to general availability.