Mistral Releases Magistral, Its First-Ever Reasoning AI Models With Transparent Chain-of-Thought

French AI startup Mistral released two reasoning-focused AI models: Magistral Small (open-source, 24B parameters) and Magistral Medium (enterprise, proprietary). Both feature multi-step reasoning with transparent chain-of-thought in eight languages. Small scored 70.7% on AIME2024, Medium achieved 73.6%. The models target research, strategic planning, and data-driven decisions, available via Hugging Face and cloud platforms.READ FULL