Magistral Medium — benchmark results

Mistral's first frontier-class reasoning model, the larger API-only counterpart to Magistral Small (June 2025). Provider: Mistral. Released 2025-06-10. Access: API.

Unified ELO 1553 ± 19, rank #613 of 1776 rated models, from 85 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
SpeechMap Compliance92.6% Requests Completed93.8
HumainE Leaderboard30.88HumainE Score73.6
AA MATH-50091.73Accuracy (%)68.8
AA Omniscience - Software Engineering (SWE) - Rust58Accuracy (%)68.8
AA Omniscience - Science, Engineering & Mathematics30.9Accuracy (%)68.4
AA Omniscience-26.42Score66.5
AA Omniscience - Software Engineering (SWE) - Swift44Accuracy (%)63.3
AI for Education Pedagogy - Social studies80.91Accuracy (%)63.2
AA Omniscience - Health22.1Accuracy (%)62.5
AA Omniscience - Business19.2Accuracy (%)61.5
AA Humanity's Last Exam9.55Accuracy (%)59.9
AA LiveCodeBench52.7Pass@1 (%)59.4

Interactive version: theaggregate.ai/model?slug=magistral-medium · How the rankings work · Data refreshed daily, snapshot 2026-07-22.