Gemini 3.1 Pro vs Qwen3 8
Gemini 3.1 Pro 
Qwen3 8 Verdict
- Gemini 3.1 Pro and Qwen3 8 (max) are level on quality (63.9 vs 63.9).
- Gemini 3.1 Pro is stronger in agents & tools, instruction following, knowledge, long context, maths, reasoning.
- Qwen3 8 (max) is stronger in coding, composite, human preference, multimodal.
- Qwen3 8 (max) is 1.7× cheaper ($2.67 vs $4.50 per 1M blended).
- Gemini 3.1 Pro streams 2.8× faster (110 vs 39 tokens per second).
| Metric | Gemini 3.1 Pro | Qwen3 8 (max) |
|---|---|---|
| BenchLeader Index | 63.9 | 63.9 |
| Agents & tools score | 63.7 | 56.1 |
| Coding score | 59.8 | 71.7 |
| Composite score | 67.2 | 70.8 |
| Human preference score | 59.6 | 67.9 |
| Instruction following score | 69.5 | – |
| Knowledge score | 66.1 | 62.3 |
| Long context score | 67.6 | 65.7 |
| Maths score | 60.7 | – |
| Multimodal score | 66.1 | 67.2 |
| Reasoning score | 69.5 | 67.7 |
| Blended price $/M | $4.50 | $2.67 |
| Output speed | 110 tok/s | 39 tok/s |
| Time to first answer | 30.5 s | 53.2 s |
| Context window | 1.0M | 1M |
| GPQA Diamond | 94.1% | – |
| FrontierMath Tiers 1–3 | 59.6% | – |
| FrontierMath Tier 4 | 26.8% | – |
| OTIS Mock AIME | 95.6% | – |
| Humanity's Last Exam | 46.4% | – |
| Terminal-Bench | 80.2% | – |
| SimpleBench | 79.6% | – |
| SciCode | 58.9% | – |
| WeirdML | 72.1% | – |
| APEX-Agents | 33.5% | – |
| ProofBench | 26.0% | – |
| GSO-Bench | 22.6% | – |
| Epoch Capabilities Index | 155 | 156.6 |
| LMArena Text | 1487 | 1480 |
| LMArena Hard Prompts | 1507 | 1502 |
| LMArena Coding | 1521 | 1520 |
| LMArena WebDev | 1447 | 1685 |
| LMArena Vision | 1295 | 1313 |
| LMArena Agent | -5.3 | 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 | 30.4 | 40.3 |
| IFBench | 77.1% | – |
| AA-LCR | 82.0% | 78.3% |
| MMMU-Pro | 82.4% | 82.3% |
| AA-Omniscience | 31.9 | 3.4 |
| Terminal-Bench Hard | 53.8% | – |
| GPQA Diamond (AA) | 94.1% | 92.7% |
| Humanity's Last Exam (AA) | 47.0% | 43.0% |
| SciCode (AA) | 58.7% | 53.2% |
| τ²-Bench Telecom (AA) | 95.6% | – |
| 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 |
| AIME 2026 | 98.3% | – |
| HMMT February 2026 | 94.7% | – |
| MathArena Apex | 60.9% | – |
| MultiChallenge | 71.4% | – |
| PRBench Finance | 41.9% | – |
| PRBench Legal | 44.0% | – |
| MultiNRC | 64.7% | – |
| HiL-Bench | 35.3% | – |
| TutorBench | 53.0% | – |
| EQ-Bench 4 | 1142 | – |
| ARC-AGI-1 | 98.0% | – |
| ARC-AGI-2 | 77.1% | – |
| ARC-AGI-3 | 0.4% | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.