Magistral
Overview
Magistral is a language model developed by Mistral AI, designed to perform transparent, multi-step reasoning across complex tasks in various domains such as law, finance, healthcare, logistics, and software development. It supports multi-step chain-of-thought generation in multiple languages, including Arabic, Chinese, French, German, and Spanish. The model is available in two versions: a 24-billion parameter 'Small' variant, which is open-weight and can be self-hosted under the Apache 2.0 license, and a higher-end 'Medium' variant, accessible via Mistral's API, SageMaker, and other cloud platforms. The 'Medium' version is priced at $2 per million input tokens and $5 per million output tokens. Notably, the 'Small' version offers a context window of 128,000 tokens, enabling it to handle extensive input sequences effectively. This design allows Magistral to deliver step-by-step answers across multiple languages, making it suitable for applications requiring transparent logic for calculations, coding, or regulated-industry workflows. Its ability to expose its chain-of-thought reasoning facilitates auditing and verification of each step, enhancing trust and reliability in its outputs. In benchmark evaluations, Magistral Medium achieved a score of 73.6% on the AIME 2024 benchmark, and 89.8% on the HumanEval coding benchmark, demonstrating its competitive performance in both reasoning and coding tasks. These attributes position Magistral as a robust tool for tasks demanding high-fidelity reasoning and multilingual support, distinguishing it from other models in the market.