| Benchmark | Gemini 2.5 Pro | Muse Spark |
|---|---|---|
| AA Agentic Index | 7.2 | 28.7 |
| AA Intelligence | 27 | 44.3 |
| AA-LCR | 66 | 77 |
| AA-Omniscience | -14.3 | 7.2 |
| ARC-AGI-2 | 4.9 | 42.5 |
| arena_elo | 1446 | 1488 |
| arena_text_factuality | 1462 | 1467 |
| arena_vision | 1246 | 1295 |
| Artificial Analysis Coding Index | 46.7 | 58.6 |
| coding_arena_elo | 1224 | 1540 |
| critpt | 2.6 | 11.3 |
| Fortress | 54.9 | 20.2 |
| frontiermath_tier_4 | 4.2 | 14.6 |
| gdpval | 8.5 | 32.3 |
| GPQA Diamond | 84.4 | 89.5 |
| HLE | 22.5 | 58.4 |
| IFBench | 49 | 75.9 |
| MMMU-Pro | 74.9 | 80.5 |
| multichallenge | 53.6 | 75.5 |
| OmniScience Accuracy | 39 | 49.6 |
| OmniScience Non-Hallucination | 12.6 | 26.8 |
| scicode | 43 | 51.5 |
| simpleqa_verified | 56 | 66.3 |
| SWE-bench Verified | 67.2 | 77.4 |
| TauBench V3 - Banking | 9.7 | 19.6 |
| Terminal-Bench 2.1 | 28.5 | 62.2 |
| Terminal-Bench Hard | 26.5 | 45.5 |
| τ²-Bench Telecom (AA run) | 54.1 | 91.5 |
| τ³-Bench | 9.3 | 19.6 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.