Up to four models
MiniMax-M2.7 vs Mistral Medium 3.5
On the index (GPT-4 = 100), MiniMax-M2.7 scores 132 and Mistral Medium 3.5 131. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
132114–151
WeightsOpen
ReleasedMar 2026
Price in / out$0.21 / $0.84
Context197K
Core tests taken1 of 9
131114–148
WeightsOpen
ReleasedApr 2026
Price in / out$1.5 / $7.5
Context262K
Core tests taken1 of 9
MiniMax-M2.7 and Mistral Medium 3.5 are too close to call (1 point apart, within the uncertainty), ahead on 0 of 1 shared core tests, and costs 8.2× less.
Test by testthird-party results
Every result as a table
| Test | MiniMax-M2.7 | Mistral Medium 3.5 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Vals AI) | 86.6% | 34.8% |
| MMLU Pro | 80.4% | 75.3% |
| Coding | ||
| Code Migration | 8.7% | 5.1% |
| SWE-bench | 73.8% | 66.4% |
| Vibe Code Bench v1.1 | 11.9% | 2.9% |
| WeirdML | 37.0% | 43.7% |
| Professional work | ||
| Finance Agent (v2) | 27.9% | 32.1% |
| CorpFin v2 | 61.2% | 58.8% |
| EMB | 28.4% | 11.2% |
| Legal Research Bench | 10.6% | 9.1% |
| MedCode | 34.4% | 33.8% |
| MedScribe | 79.9% | 67.7% |
| Tax Agent Bench | 6.9% | 4.5% |
| TaxEval v2 | 66.6% | 68.0% |
Tests at least two of these models have taken. Bold rows are the core tests.