Frontier Log

Mistral

Magistral Small

Announced on mistral.ai

Benchmark results

BenchmarkScoreSource check
AIME 2024No tools · pass@170.7%
CorroboratedThe same number is printed in 1 other lab document: 1
majority vote over 64 samples · parallel test-time compute83.3%
AIME 2025No tools · pass@162.76%
majority vote over 64 samples · parallel test-time compute · in Magistral76.7%
LiveCodeBenchLiveCodeBench v555.84%
LiveCodeBench v6 · in Magistral47.4%
Sources

Reported by Mistral

Related events

All models