Mistral unveils Mistral Large 4 “Le Chonk”, a new frontier in open-weight AI models

French AI company Mistral has officially announced the public preview of its latest model, Mistral Large 4, affectionately dubbed “Le Chonk”. The new model represents a significant leap forward in open-weight AI performance, boasting a massive 1 trillion total parameters with 49 billion active parameters in its native mixture-of-experts architecture. As of October 6, 2026, developers can access the model through Mistral Studio's API while the company prepares to release full weights by the end of the month. The model is natively multimodal and fluent in over 160 languages, unifying instruction-following, reasoning, and agentic capabilities in a single system.

According to Mistral's announcement, the model achieves state-of-the-art performance among open-weight systems on critical enterprise workloads including cybersecurity, finance, and manufacturing. In some domains such as visual grounding, the company claims it surpasses even frontier closed models. The pricing structure reflects premium positioning at $1.36 per million tokens for input and $4.18 per million tokens for output. Until the weights are publicly released, Mistral is conducting real-world red-teaming with cybersecurity leaders, vetted partners, and state authorities who receive access with reduced moderation and expanded cyber capabilities. This strategic rollout signals Mistral's commitment to balancing innovation with responsible deployment as it positions “Le Chonk” as the foundation for a new generation of specialized industrial models.