Magistral Small — benchmark results

Mistral's Apache-2.0 open 24B reasoning model, SFT-distilled from Magistral Medium traces plus RL (June 2025). Provider: Mistral. Released 2025-06-10. Access: Open.

Unified ELO 1546 ± 10, rank #633 of 1776 rated models, from 284 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Swedish NLU - Multi Wiki QA SV82.4Reading comprehension Score (%)99.6
EuroEval Slovak NLU - Multi Wiki QA SK74.29Reading comprehension Score (%)99
EuroEval Danish NLU - Multi Wiki QA DA82.72Reading comprehension Score (%)98.9
EuroEval Faroese NLU - FoQA79.56Reading comprehension Score (%)98.7
EuroEval Italian NLU - SQuAD IT76.71Reading comprehension Score (%)98.6
EuroEval Norwegian NLU - Norquad77.06Reading comprehension Score (%)98.4
EuroEval Latvian NLU - MultiWikiQA LV71.87Reading comprehension Score (%)97.8
EuroEval English77.21Average Score (%)97.6
EuroEval Lithuanian NLU - MultiWikiQA LT73.58Reading comprehension Score (%)97.4
EuroEval Spanish NLU - MLQA ES69.43Reading comprehension Score (%)97.3
EuroEval English NLU - SQuAD88.59Reading comprehension Score (%)97.2
EuroEval Slovak NLU61.56NLU Average Score (%)97

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