Compare / Claude Opus 4.8 vs MiniMax M3
Claude Opus 4.8 vs MiniMax M3
20 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Claude Opus 4.8 | MiniMax M3 |
|---|---|---|---|
| APEX-Agentsreported | Tasks passed% of tasks | 48.9% | 37.7% |
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | 0.04 | -0.14 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | 0.15 | -0.13 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | 0.10 | -0.07 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | -0.00 | -0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,486 | 1,433 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,469 | 1,434 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,490 | 1,450 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,467 | 1,406 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,525 | 1,471 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,491 | 1,435 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,480 | 1,440 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,474 | 1,422 |
| OpenHands Indexreported | Average% resolved | 71.9% | 57.2% |
| OpenHands Indexreported | Front end (SWE-Bench Multimodal)% resolved | 50.0% | 36.8% |
| OpenHands Indexreported | Greenfield (Commit0)% resolved | 62.5% | 25.0% |
| OpenHands Indexreported | Information gathering (GAIA)% resolved | 78.8% | 66.7% |
| OpenHands Indexreported | Issue resolution (SWE-Bench)% resolved | 83.8% | 76.4% |
| OpenHands Indexreported | Testing (SWT-Bench)% resolved | 84.3% | 81.1% |
| Vending-Bench 2reported | Money after a yearUS dollars | $5,787.43 | $2,157.77 |
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.