Compare / Qwen3.5 397B A17B vs gemini-3-flash
Qwen3.5 397B A17B vs gemini-3-flash
21 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Qwen3.5 397B A17B | gemini-3-flash |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,445 | 1,468 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,445 | 1,469 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,405 | 1,457 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,476 | 1,496 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,434 | 1,458 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,442 | 1,473 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,421 | 1,461 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 21.4 | 26.3 |
| Artificial Analysisreported | GPQA Diamond% of questions | 89.3% | 89.8% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 29.0% | 36.6% |
| Artificial Analysisreported | IFBench% of instructions | 78.8% | 78.0% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 77.3% | 78.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 40.9% | 38.6% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 13.4% | 20.8% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 95.6% | 80.4% |
| tau2-benchreported | Consistency (pass^4)% of tasks | 68.0% | 68.0% |
| tau2-benchreported | Task success (pass^1)% of tasks | 81.5% | 82.5% |
| tau2-benchreported | Consistency (pass^4)% of tasks | 59.6% | 51.8% |
| tau2-benchreported | Task success (pass^1)% of tasks | 84.4% | 76.8% |
| tau2-benchreported | Consistency (pass^4)% of tasks | 92.1% | 70.2% |
| tau2-benchreported | Task success (pass^1)% of tasks | 97.8% | 91.2% |
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.