Magistral Medium 1 is Mistral's mid-tier reasoning model from June 2025. It edges out Magistral Small on most benchmarks, with a GPQA of 0.679, HLE of 0.095, and livecodebench of 0.527, placing it in a competitive range for mid-size models. However, its agentic index of 16.1 and terminalbench score of 0.09 confirm it is not built for autonomous, multi-step workflows.

For business use, it fits structured coding tasks, technical Q&A, and moderate-complexity content generation. Long-context reliability and instruction-following are not standout strengths, so it is less suited to document-heavy or precision-critical workflows. It performs better than the Small variant on scientific reasoning tasks.

Without published pricing, cost comparisons are difficult, but Mistral's mid-tier models are generally competitively priced. Teams needing a step up from Magistral Small without committing to a frontier model will find this a reasonable option for technical workloads.

Assessed June 30, 2026

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?