Mistral Medium: benchmark results
Mistral's original Mistral Medium, an API-only model released December 2023, the first entry in what became their ongoing Medium product line. Provider: Mistral. Released 2023-12-11. Access: API.
Aggregate Rating 1497 ± 14, rank #792 of 1656 rated models, from 88 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AlpacaEval | 91.54 | Length-Controlled Win Rate (%) | 98.2 |
| AlpacaEval 1.0 | 96.83 | Win Rate (%) | 98 |
| SpeechMap Compliance | 91.7 | % Requests Completed | 93.8 |
| EvalYaks CEFR B2 (With Descriptors) | 80 | Average acceptable accuracy (%) across speaking-test parts 1 | 90 |
| BiGGen-Bench | 3.94 | Average Score (1-5) | 89.2 |
| MAGI-Hard | 62.15 | Accuracy (%, 2024-05 snapshot) | 75.9 |
| BenchBench | 65.71 | Aggregate Score (%) | 70.6 |
| AlpacaEval 2.0 | 28.61 | LC Win Rate (%) | 66.2 |
| InfiBench | 40.95 | Score (%) | 59 |
| MixEval | 47.8 | Score | 56.9 |
| HELM Lite - LegalBench - Abercrombie | 61.05 | Exact Match (%) | 56.1 |
| WMT25 English-Bhojpuri chrF++ | 31.6 | chrF++ | 52.9 |
Interactive version: theaggregate.ai/model?slug=mistral-medium · How It Works · Data refreshed daily, snapshot 2026-10-20.