Mistral unveils ‘Le Chonk’ to compete with Chinese open-weight AI models
The open weights for Le Chonk are set to release at the end of October, announced Mistral
On Tuesday, October 6, 2026, French AI startup Mistral unveiled its new AI model, Le Chonk, aiming to compete against the leading open-source models. The company described Le Chonk as its "largest and most capable model to date" with 1 trillion parameters and 49 billion active parameters. In comparison, the Chinese AI company Moonshot's Kimi K3 model, launched in July, has 2.8 trillion parameters.
CEO Artur Mensch stated that Le Chonk outperforms open-weight models from the US or Europe, particularly in areas like coding, agentic workflows, and multimodal understanding. With fluency in over 160 languages, Le Chonk excels in cybersecurity, finance, law, and visual grounding. The French startup claims that the new model is a counterweight to the closed frontier models in the U.S., which have been criticized for accessibility issues and security concerns.
Mistral trained Le Chonk from scratch on 3,800 NVIDIA Grace Blackwell GPUs in its European data centers. The model's open weights are expected to be released by the end of October, currently available as a public preview. Mistral stated that they will red-team the model in real-world settings with cybersecurity leaders, vetted partners, and state authorities, who will access the model with reduced moderation and expanded cyber capabilities.
The company announced that users can expect a new generation of specialized and optimized Mistral models in the future, emphasizing a fast pace of progress. This stance contrasts with the messages from U.S. AI company CEOs, who have called for a more cautious approach to AI development due to concerns over its potential implications.
Written by urgent.news from The Hindu - Sci-Tech's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
