Mistral says "Le Chonk" can challenge the best AI models

French AI company Mistral has released a new freely available model called Mistral Large 4, nicknamed Le Chonk, which it says can compete with the best models from the US and China. The 1 trillion-parameter model is currently available in preview, with a final version expected by the end of the month. According to Mistral, it can be used and customized by anyone.

Le Chonk is designed as a general-purpose model but is specifically optimized for coding and cyberdefense, as well as tasks in manufacturing, finance, electrical engineering, and other niches. Guillaume Lample, Mistral's cofounder and chief scientist, told WIRED that other labs do not focus much on these areas. Mistral presents Le Chonk as by far the most capable open-weight model developed outside China and "very, very close" to some proprietary models.

Mistral claims to have trained the model from scratch, distinguishing it from Chinese labs that the US government has accused of abusing distillation—training a smaller model on the outputs of a larger one—to close the performance gap with OpenAI and Anthropic. Open-weight models are considerably cheaper for businesses to run, costing only as much as the compute they consume. Mistral says Le Chonk will eliminate the few remaining reasons a business might hesitate to choose open source by reducing the performance gap with leading proprietary models and offering a competitive alternative to Chinese releases.

Mistral has less capital and fewer compute resources than OpenAI and Anthropic and has generally lagged on model performance, revenue, and release frequency. It generates revenue through pay-as-you-go fees for running models via its cloud and by deploying engineers to help customers tune models. In September, Mistral raised $3.3 billion at a $24 billion valuation, the largest ever raise by a European tech company, and its earnings have reportedly increased 20-fold in the last year or so. The launch comes amid US government restrictions on model distribution and growing tensions over access to frontier AI.

Why it matters: Businesses concerned about sudden loss of access to proprietary models now have a freely available, customizable alternative optimized for coding, cyberdefense, and other domains. Mistral argues that relying on a closed model risks the collapse of defenses if access is revoked.