Gemini 3.1 Pro vs Muse Spark
Gemini 3.1 Pro 
Muse Spark Verdict
- Muse Spark leads on quality: 65.4 vs 63.9.
- Gemini 3.1 Pro is stronger in knowledge, long context, maths, multimodal.
- Muse Spark is stronger in agents & tools, coding, composite, human preference, instruction following, reasoning.
| Metric | Gemini 3.1 Pro | Muse Spark |
|---|---|---|
| BenchLeader Index | 63.9 | 65.4 |
| Agents & tools score | 63.7 | 66.3 |
| Coding score | 59.8 | 66.0 |
| Composite score | 67.2 | 68.4 |
| Human preference score | 59.6 | 69.0 |
| Instruction following score | 69.5 | 79.6 |
| Knowledge score | 66.1 | 64.5 |
| Long context score | 67.6 | 65.5 |
| Maths score | 60.7 | 56.3 |
| Multimodal score | 66.1 | 65.9 |
| Reasoning score | 69.5 | 70.0 |
| Blended price $/M | $4.50 | – |
| Output speed | 110 tok/s | – |
| Time to first answer | 30.5 s | – |
| Context window | 1.0M | 262k |
| GPQA Diamond | 94.1% | 89.8% |
| FrontierMath Tiers 1–3 | 59.6% | – |
| FrontierMath Tier 4 | 26.8% | – |
| OTIS Mock AIME | 95.6% | 88.9% |
| Humanity's Last Exam | 46.4% | 40.6% |
| Terminal-Bench | 80.2% | – |
| SimpleBench | 79.6% | – |
| SciCode | 58.9% | 51.5% |
| WeirdML | 72.1% | – |
| APEX-Agents | 33.5% | – |
| ProofBench | 26.0% | 17.0% |
| GSO-Bench | 22.6% | – |
| Epoch Capabilities Index | 155 | 152.1 |
| LMArena Text | 1487 | 1488 |
| LMArena Hard Prompts | 1507 | 1505 |
| LMArena Coding | 1521 | 1526 |
| LMArena WebDev | 1447 | – |
| LMArena Vision | 1295 | 1306 |
| LMArena Agent | -5.3 | – |
| AA Intelligence Index | 30.4 | 31.3 |
| IFBench | 77.1% | 75.9% |
| AA-LCR | 82.0% | 78.0% |
| MMMU-Pro | 82.4% | 80.5% |
| AA-Omniscience | 31.9 | 7.2 |
| Terminal-Bench Hard | 53.8% | 45.5% |
| GPQA Diamond (AA) | 94.1% | 88.4% |
| Humanity's Last Exam (AA) | 47.0% | 40.7% |
| SciCode (AA) | 58.7% | – |
| τ²-Bench Telecom (AA) | 95.6% | 91.5% |
| AIME (Vals) | – | 96.9% |
| MMLU-Pro | – | 87.3% |
| LegalBench | – | 84.2% |
| CorpFin | – | 65.1% |
| TaxEval | – | 77.7% |
| SWE-bench (Vals) | – | 74.4% |
| GPQA Diamond (Vals) | – | 89.7% |
| AIME 2026 | 98.3% | – |
| HMMT February 2026 | 94.7% | – |
| MathArena Apex | 60.9% | – |
| SWE-Bench Pro | – | 55.0% |
| MCP Atlas | – | 82.2% |
| MultiChallenge | 71.4% | 75.5% |
| PRBench Finance | 41.9% | 52.4% |
| PRBench Legal | 44.0% | 52.3% |
| MultiNRC | 64.7% | 59.0% |
| HiL-Bench | 35.3% | – |
| TutorBench | 53.0% | 68.5% |
| 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.