Compare / Mistral Medium 3.5 vs GPT-5.4 Mini
Mistral Medium 3.5 vs GPT-5.4 Mini
29 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-5.4 Mini |
|---|---|---|---|
| BulletBenchmeasured by us | Cost per gameUS dollars | $0.0354 | $0.0205 |
| BulletBenchmeasured by us | Games lost on time% of games | 0.0% | 50.0% |
| BulletBenchmeasured by us | Invalid moves% of moves | 0.4% | 0.0% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 572 | 197 |
| BulletBenchmeasured by us | Median move timemilliseconds | 0.5 s | 0.9 s |
| BulletBenchmeasured by us | Cost per gameUS dollars | $0.0297 | $0.0136 |
| BulletBenchmeasured by us | Games lost on time% of games | 0.0% | 0.0% |
| BulletBenchmeasured by us | Invalid moves% of moves | 0.0% | 0.0% |
| BulletBenchmeasured by us | Ladder Eloladder Elo | 529 | 571 |
| BulletBenchmeasured by us | Median move timemilliseconds | 0.5 s | 0.9 s |
| Arena (formerly LMArena)reported | OverallArena rating | 1,430 | 1,447 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,430 | 1,460 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,398 | 1,402 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,434 | 1,480 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,421 | 1,434 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,426 | 1,447 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,404 | 1,424 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 46.9 | 56.1 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 14.2 | 24.1 |
| Artificial Analysisreported | GPQA Diamond% of questions | 74.8% | 87.5% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 13.8% | 28.1% |
| Artificial Analysisreported | IFBench% of instructions | 68.8% | 73.3% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 69.3% | 77.0% |
| Artificial Analysisreported | SciCode% of problems | 40.2% | 52.1% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 50.6% | 59.2% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 0.0% | 2.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 33.3% | 52.3% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 15.1% | 25.6% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 94.2% | 83.3% |
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.