Compare / Claude Sonnet 4.6 vs gemini-3-pro
Claude Sonnet 4.6 vs gemini-3-pro
22 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Claude Sonnet 4.6 | gemini-3-pro |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,478 | 1,451 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,201 | 1,201 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,461 | 1,481 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,482 | 1,474 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,449 | 1,484 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,506 | 1,498 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,475 | 1,473 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,472 | 1,485 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,455 | 1,481 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 30.1 | 28.0 |
| Artificial Analysisreported | GPQA Diamond% of questions | 87.5% | 90.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 33.6% | 39.7% |
| Artificial Analysisreported | IFBench% of instructions | 56.6% | 70.4% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 80.0% | 76.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 53.0% | 41.7% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 79.5% | 87.1% |
| OpenHands Indexreported | Average% resolved | 44.5% | 49.0% |
| OpenHands Indexreported | Front end (SWE-Bench Multimodal)% resolved | 30.9% | 36.8% |
| OpenHands Indexreported | Greenfield (Commit0)% resolved | 50.0% | 25.0% |
| OpenHands Indexreported | Information gathering (GAIA)% resolved | 13.3% | 44.2% |
| OpenHands Indexreported | Issue resolution (SWE-Bench)% resolved | 74.4% | 70.6% |
| OpenHands Indexreported | Testing (SWT-Bench)% resolved | 54.0% | 68.6% |
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.