Compare / Mistral Medium 3.5 vs GPT-5.6 Sol
Mistral Medium 3.5 vs GPT-5.6 Sol
26 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Mistral Medium 3.5 | GPT-5.6 Sol |
|---|---|---|---|
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | -0.16 | 0.03 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | -0.15 | 0.19 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | -0.07 | 0.07 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.02 | 0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,430 | 1,476 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,430 | 1,485 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,398 | 1,470 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,434 | 1,534 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,421 | 1,489 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,426 | 1,483 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,404 | 1,483 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 46.9 | 78.3 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 14.2 | 47.0 |
| Artificial Analysisreported | GPQA Diamond% of questions | 74.8% | 94.1% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 13.8% | 49.5% |
| Artificial Analysisreported | IFBench% of instructions | 68.8% | 72.7% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 69.3% | 84.0% |
| Artificial Analysisreported | SciCode% of problems | 40.2% | 57.8% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 50.6% | 89.5% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.0% | 39.9% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 33.3% | 65.9% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 15.1% | 44.3% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 94.2% | 85.1% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 18.0% | 1.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.35 | 0.37 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 45.5 | 65.4 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.