Mistral AI
Mistral Medium 3 benchmark results
8 results on 7 benchmarks from 3 sources.
Closed Released: May 7, 2025 Updated Oct 22, 2026
Also listed as: mistral-medium-2505 · Mistral-Medium-2505 · Mistral-Medium-2505 (FC)
All benchmark results
| Benchmarks | Score | Rank | Setting | Source type | Date |
|---|---|---|---|---|---|
| LMArena Text | 1387 | 179 | Overall, style control on; 32786 votes; 95% CI 1383–1392“mistral-medium-2505” | Official leaderboard | Oct 8, 2026 |
| LMArena Vision | 1155 | 109 | Overall, style control on; 10809 votes; 95% CI 1147–1163“mistral-medium-2505” | Official leaderboard | Oct 2, 2026 |
| GPQA Diamond | 59.5% | 183 | Epoch AI run, accuracy | Independent run | May 7, 2025 |
| AIME (OTIS Mock AIME 2024–2025) | 32.2% | 189 | Epoch AI run, OTIS Mock AIME 2024–2025 | Independent run | May 7, 2025 |
| MATH Level 5 | 81.6% | 36 | Epoch AI run, accuracy | Independent run | May 7, 2025 |
| Humanity's Last Exam | 4.5% | 47 | Compiled by Epoch AI | Third-party compilation via Epoch AI | May 7, 2025 |
| BFCL v4 (function calling) | 37.7% | 48 | Overall accuracy“Mistral-Medium-2505” | Official leaderboard | Oct 17, 2026 |
| BFCL v4 (function calling) | 37.6% | 49 | Overall accuracy; native function calling (FC)“Mistral-Medium-2505 (FC)” | Official leaderboard | Oct 17, 2026 |
FAQ
Where do these Mistral Medium 3 scores come from?
From Epoch AI Benchmarking Hub, Berkeley Function Calling Leaderboard (BFCL), and LMArena leaderboard dataset. Each row links to the original leaderboard, dataset or report, and vendor-reported numbers are labelled.