Mistral releases updated Le Chonk model for coding and cybersecurity

In short: Mistral says its updated AI model, Le Chonk, is now its best open-weight tool for coding and cybersecurity, and anyone can access it.

What happened

French AI company Mistral has released a finalized version of its model nicknamed Le Chonk. The company says the updated model performs especially well at automated coding and cybersecurity work.

Mistral describes Le Chonk as an “open-weight” model. That means the core files that make it work, called weights (think of them like the model’s saved “memory” and skills), are available for others to download and run. Mistral says this makes it harder for any one company to limit access later.

Mistral says it improved Le Chonk by training it on harder and harder coding and security problems. It also says the model, which it describes as “trillion-parameter” (a measure of size and complexity), can match or beat models from Chinese labs such as DeepSeek and Z.ai on these tasks.

Why it matters

Cyberattacks are rising, and AI tools are being used on both sides, by defenders and by attackers. Mistral argues that businesses and governments in the West need strong open-weight models they can run themselves, instead of relying only on closed systems that could be turned off remotely.

Security experts also warn there is a trade-off. If a powerful model is free to download, criminals can download it too and customize it for harmful use. One security researcher told WIRED the bigger issue is not the model’s country of origin, but whether organizations put strong monitoring and controls in place.

Source: Wired