Mistral AI Launches Large 4 With 1 Trillion Parameters as European AI Alternative
French AI startup Mistral AI released a public preview of its Mistral Large 4 model on October 6, 2026. The model has approximately 1 trillion parameters and uses a sparse Mixture-of-Experts architecture. Mistral plans to release the model weights publicly on October 27, positioning it as a sovereign European alternative to U.S. and Chinese AI systems.
French AI startup Mistral AI released a public preview of its Mistral Large 4 model on October 6, 2026. The model, nicknamed "Le Chonk" internally, has approximately 1.05 trillion parameters and uses a sparse Mixture-of-Experts architecture that activates about 49 to 52 billion parameters per token during inference.
Mistral plans to release the model weights publicly on October 27, 2026, following a period of safety testing. The company is positioning Large 4 as a sovereign European alternative to frontier models from U.S. companies like OpenAI and Anthropic, and from Chinese developers.
The model is natively multimodal, meaning it was trained from the start to process both text and images. It includes a 1.6-billion-parameter vision encoder and supports a context window of 1 million tokens. Mistral trained the model on roughly 3,800 to 4,000 Nvidia Grace Blackwell GPUs in its European data centers.
On performance benchmarks, Mistral reported 61.7 percent on DeepSWE v1.1 for coding tasks and 93 percent on the Cybench cybersecurity suite. The company says the model is designed for enterprise and institutional customers in defense, finance, and chip design who need to host AI locally for data security and auditability.
The model is currently available in preview through the Mistral Studio API. Pricing is set at approximately 1.36 dollars per million input tokens and 4.18 dollars per million output tokens.
