Mistral Large 4

Mistral Large 4 1T-A52B is Mistral AI's flagship open-weight multimodal Mixture-of-Experts model, designed for advanced reasoning, coding, visual understanding, and autonomous agents. It contains approximately one trillion total parameters, with 49B active per token in its backbone and 52B including embeddings and output layers. Its architecture uses 128 routed experts, four active per token, and one shared expert. The model supports a one-million-token context window and accepts text and images while generating text. It offers configurable low, medium, and high reasoning effort, native function calling, and strict JSON schema enforcement. Fluent in more than 160 languages, it targets software engineering, cybersecurity, finance, law, and complex enterprise workflows. An integrated DSpark speculative decoding head accelerates inference, while deployment and fine-tuning are supported through vLLM, SGLang, Transformers, and other frameworks.