Gemini 3.1 Pro vs GPT-5.6 Sol
Gemini 3.1 Pro 
GPT-5.6 Sol Verdict
- GPT-5.6 Sol (high) leads on quality: 68.0 vs 63.9.
- Gemini 3.1 Pro is stronger in human preference, instruction following, long context, maths, reasoning.
- GPT-5.6 Sol (high) is stronger in agents & tools, coding, composite, knowledge, multimodal.
- Gemini 3.1 Pro is 1.8× cheaper ($4.50 vs $8.00 per 1M blended).
- Gemini 3.1 Pro streams 1.9× faster (110 vs 59 tokens per second).
| Metric | Gemini 3.1 Pro | GPT-5.6 Sol (high) |
|---|---|---|
| BenchLeader Index | 63.9 | 68.0 |
| Agents & tools score | 63.7 | 87.6 |
| Coding score | 59.8 | 72.0 |
| Composite score | 67.2 | 82.5 |
| Human preference score | 59.6 | – |
| Instruction following score | 69.5 | 67.6 |
| Knowledge score | 66.1 | 73.7 |
| Long context score | 67.6 | 67.4 |
| Maths score | 60.7 | – |
| Multimodal score | 66.1 | 66.5 |
| Reasoning score | 69.5 | 62.9 |
| Blended price $/M | $4.50 | $8.00 |
| Output speed | 110 tok/s | 59 tok/s |
| Time to first answer | 30.5 s | 31.4 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% | 56.9% |
| WeirdML | 72.1% | 88.8% |
| APEX-Agents | 33.5% | – |
| ProofBench | 26.0% | – |
| GSO-Bench | 22.6% | – |
| Epoch Capabilities Index | 155 | – |
| LMArena Text | 1487 | – |
| LMArena Hard Prompts | 1507 | – |
| LMArena Coding | 1521 | – |
| LMArena WebDev | 1447 | – |
| LMArena Vision | 1295 | – |
| LMArena Agent | -5.3 | – |
| AA Intelligence Index | 30.4 | 42.5 |
| IFBench | 77.1% | 69.2% |
| AA-LCR | 82.0% | 81.7% |
| MMMU-Pro | 82.4% | 81.8% |
| AA-Omniscience | 31.9 | 20.4 |
| Terminal-Bench Hard | 53.8% | 62.1% |
| GPQA Diamond (AA) | 94.1% | 92.8% |
| Humanity's Last Exam (AA) | 47.0% | 46.0% |
| SciCode (AA) | 58.7% | 57.8% |
| τ²-Bench Telecom (AA) | 95.6% | 83.3% |
| AIME 2026 | 98.3% | – |
| HMMT February 2026 | 94.7% | – |
| MathArena Apex | 60.9% | – |
| MultiChallenge | 71.4% | – |
| PRBench Finance | 41.9% | – |
| PRBench Legal | 44.0% | – |
| EnigmaEval | – | 37.1% |
| MultiNRC | 64.7% | – |
| HiL-Bench | 35.3% | – |
| TutorBench | 53.0% | – |
| EQ-Bench 4 | 1142 | – |
| ARC-AGI-1 | 98.0% | 97.0% |
| ARC-AGI-2 | 77.1% | 85.4% |
| ARC-AGI-3 | 0.4% | 2.2% |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.