Mistral Large 2411 is an advanced dense large language model built with 123 billion parameters, representing Mistral AI's continued refinement of its flagship reasoning and coding line. Designed to be agent-centric from the ground up, it brings best-in-class function calling, JSON output structuring, and robust context adherence for retrieval-augmented generation and other large-context workflows. The model was trained on over 80 programming languages including Python, Java, C, C++, JavaScript, Swift, and Fortran, positioning it as a strong choice for software engineering tasks. Its multilingual design natively supports dozens of languages, and the expanded 131K token context window enables it to hold complex, multi-turn conversations and process lengthy documents with strong adherence to source material.

As an update to the earlier Mistral Large 2 release, Mistral Large 2411 arrived in November 2024 alongside Pixtral Large 2411, bringing notable improvements in long-context comprehension, system prompt handling, and function-calling accuracy over the previous 2407 iteration. The model was trained with a knowledge cutoff through mid-2024 and continues to emphasize state-of-the-art reasoning and mathematical capabilities as core strengths. Practitioners value it for its balance of structured output reliability, coding proficiency, and the ability to follow complex instructions reliably within extended conversations. These qualities make it particularly well-suited for building autonomous agents, automated pipelines, and applications that demand consistent, high-quality generation across diverse use cases.