French AI lab Mistral AI released Mistral Large 4, a new 1-trillion-parameter multimodal model named Le Chonk, aiming to leapfrog both American and Chinese rivals. The model was unveiled on Tuesday by the company, positioning itself as a European alternative amid the growing divide between closed and open AI models. While initially accessible only via a public guardrail endpoint, Mistral plans to make the open-source weights available in three weeks following safety testing.
VP of Science Pierre Stock explained that the model was trained entirely on Mistral's compute using only 4,000 Nvidia GPUs, which is significantly less compute than its competitors use. The model was trained with a focus on optimized use cases for enterprise customers, including cybersecurity, finance, and chip design. These areas align closely with key backers such as ASML and Samsung, which recently led Mistral's Series D funding round at a €21 billion valuation.
Security concerns have driven the decision to delay the release of the weights, as Mistral works with trusted partners and governments to ensure the model cannot be used for malicious attacks. Although official benchmark results are still pending, the company hopes the new model will become best-in-class among open-weight models outside of China. With this major release, Mistral aims to reinforce its status as a leading frontier lab rather than just an inference provider.
This new model matters because it strengthens Europe's competitive position in the global AI race by offering a high-capacity alternative to American and Chinese frontier systems.