Gemini 3.7 Flash vs Qwen3 8
Gemini 3.7 Flash 
Qwen3 8 Verdict
- Gemini 3.7 Flash (medium) and Qwen3 8 (max) are level on quality (64.4 vs 63.9).
- Gemini 3.7 Flash (medium) is stronger in composite, knowledge, long context, multimodal.
- Qwen3 8 (max) is stronger in coding, reasoning, agents & tools, human preference.
- Gemini 3.7 Flash (medium) is 1.8× cheaper ($1.50 vs $2.67 per 1M blended).
- Gemini 3.7 Flash (medium) streams 7.2× faster (282 vs 39 tokens per second).
| Metric | Gemini 3.7 Flash (medium) | Qwen3 8 (max) |
|---|---|---|
| BenchLeader Index | 64.4 | 63.9 |
| Coding score | 68.3 | 71.7 |
| Composite score | 78.9 | 70.8 |
| Knowledge score | 75.2 | 62.3 |
| Long context score | 68.1 | 65.7 |
| Multimodal score | 69.4 | 67.2 |
| Reasoning score | 63.5 | 67.7 |
| Agents & tools score | – | 56.1 |
| Human preference score | – | 67.9 |
| Blended price $/M | $1.50 | $2.67 |
| Output speed | 282 tok/s | 39 tok/s |
| Time to first answer | 5.3 s | 53.2 s |
| Context window | 1M | 1M |
| SciCode | 57.9% | – |
| Epoch Capabilities Index | – | 156.6 |
| LMArena Text | – | 1480 |
| LMArena Hard Prompts | – | 1502 |
| LMArena Coding | – | 1520 |
| LMArena WebDev | – | 1685 |
| LMArena Vision | – | 1313 |
| LMArena Agent | – | 3.9 |
| LiveBench | – | 78.5% |
| LiveBench Reasoning | – | 88.2% |
| LiveBench Coding | – | 72.9% |
| LiveBench Agentic Coding | – | 64.7% |
| LiveBench Mathematics | – | 91.3% |
| LiveBench Data Analysis | – | 78.4% |
| LiveBench Language | – | 79.7% |
| AA Intelligence Index | 39.6 | 40.3 |
| AA-LCR | 83.0% | 78.3% |
| MMMU-Pro | 84.7% | 82.3% |
| AA-Omniscience | 23.7 | 3.4 |
| GPQA Diamond (AA) | 92.1% | 92.7% |
| Humanity's Last Exam (AA) | 39.0% | 43.0% |
| SciCode (AA) | 59.8% | 53.2% |
| LiveCodeBench | – | 87.8% |
| MMLU-Pro | – | 88.6% |
| IOI | – | 73.0% |
| LegalBench | – | 83.6% |
| CorpFin | – | 65.8% |
| TaxEval | – | 75.5% |
| Terminal-Bench 2.1 (Vals) | – | 67.4% |
| SWE-bench (Vals) | – | 85.6% |
| GPQA Diamond (Vals) | – | 93.7% |
| Vals Index | – | 51.8 |
| ARC-AGI-1 | 91.2% | – |
| ARC-AGI-2 | 63.8% | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.