Magistral Medium
Mistral·
LLMsflagshipcloud
Overview
Mistral's first reasoning model. API and enterprise deployment only — no published weights. The 24B Magistral Small released the same day carries open weights and is tracked separately in the local catalog.
Capabilities and innovations
Enterprise focusPrivacy firstSpecialized domains
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 72.4% | unsourced |
| Humanity’s Last Exam (no tools) | 19.8% | unsourced |
| SWE-bench Verified | 48.9% | unsourced |
| MMLU | 88.4% | unsourced |
- SWE-bench Pro: not reported by the vendor
- MMLU-Pro: not reported by the vendor
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Links
More from Mistral
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.