Mistral launched a public preview of Mistral Large 4 on October 6, 2026, a 1.05-trillion-parameter Mixture-of-Experts model the company says will become the most capable open-weight system developed outside China once its weights drop later this month7,8. Mistral VP Science Pierre Stock told TechCrunch the model was trained entirely on Mistral's compute using only 4,000 NVIDIA Grace Blackwell GPUs, "which is two to three times less than our Chinese competitors, and significantly less than the closed source competitors"2,3. Mistral cofounder and chief scientist Guillaume Lample told WIRED the model targets domains where "the other labs will not focus that much"6.
The model, nicknamed Le Chonk, runs 49 billion active parameters per forward pass, includes a 1.6-billion-parameter vision encoder, and supports a one-million-token context window. It is accessible now through Mistral's API under the identifier mistral-large-4, with an open-weights release planned for October 27.
Architecture and positioning
Mistral describes Mistral Large 4 as natively multimodal, combining instruction following, reasoning, and agentic operation in a single system. The model covers more than 160 languages, including every official EU language, and works across Latin and non-Latin scripts. Supported capabilities listed in the documentation include structured outputs, function calling, document question answering, prefix support, batch processing, agents, conversations, and built-in tools.
Mistral said it trained ML4 from scratch over two months in its own European data centers. The company said the model was built in Europe end-to-end and can be deployed from Europe through Mistral Cloud infrastructure. Mistral claims it trained Le Chonk from scratch, distinguishing its approach from Chinese labs that the U.S. government has accused of using distillation from larger models to close the performance gap with OpenAI and Anthropic.
API pricing is listed at $0.68 per million input tokens, $0.07 per million cached input tokens, and $2.09 per million output tokens.
Performance claims and cyber-defense focus
Mistral said Mistral Large 4 is the best open-weights model from the United States or Europe on aggregated benchmarks and claims state-of-the-art results on critical workloads in cyber defense, manufacturing, and finance. The company said the model outperforms closed frontier models on visual grounding tasks. Mistral also claimed its cyber-defense capabilities are stronger than those of its Chinese competitors and said it is working privately with cybersecurity partners during the public preview period.
Stock said the weights are being withheld for three weeks while Mistral works "with trusted partners and governments to make sure that the open-source weights can be used to defend, but not to [perform] malicious attacks". He added that an open-weight model is easier to audit.
Competitive framing
Mistral is positioning Mistral Large 4 as a European sovereignty play in a market it frames as increasingly split between U.S. closed-model providers and Chinese open-weight leaders. CNBC reported that Mistral raised €3 billion in a September Series D at a €21 billion valuation, cementing its status as one of the best-funded AI startups in Europe5. WIRED separately reported the round at $3.3 billion at a $24 billion valuation. Mistral's earnings have reportedly increased 20-fold in the last year.
ANALYSIS At $0.68/$2.09 per million input/output tokens with 49 billion active parameters out of a 1.05-trillion-parameter MoE, Mistral is pricing Mistral Large 4 to compete on inference cost against denser frontier models while matching them on raw parameter count. The explicit cyber-defense and sovereignty messaging, combined with the staged weight release through government and partner vetting, targets European government and defense procurement channels that have limited open-weight alternatives at this capability tier.
No Mistral Large 4 checkpoint had appeared in Mistral's published Hugging Face Inc. collections as of the announcement; the most recently updated entries remain Mistral Large 3 and Mistral Small 4. The open-weights release is scheduled for October 27.