Magistral Medium 1

Magistral Medium 1 is a reasoning-focused language model developed by Mistral AI, introduced in June 2025. It represents a specialized branch of the Mistral model family designed to address advanced analytical tasks, multi-step problem solving, and complex logical reasoning. The model is categorized as a frontier-class reasoning agent, optimized for professional use cases that require higher logical consistency and deep analysis than general-purpose LLMs. Supporting a 40,000-token context window, Magistral Medium 1 is primarily utilized for enterprise-grade knowledge work, including code debugging and complex research. As a proprietary model, its specific architecture and parameter count have not been publicly disclosed, though it is offered via API in multiple version iterations including 1.0, 1.1, and 1.2.