| Benchmark | Gemini 2.5 Pro | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 7.2 | 39.7 |
| AA Intelligence | 27 | 53.2 |
| AA-LCR | 66 | 81.3 |
| AA-Omniscience | -14.3 | 28.1 |
| arena_elo | 1446 | 1491 |
| arena_text_factuality | 1462 | 1466 |
| Artificial Analysis Coding Index | 46.7 | 71.3 |
| coding_arena_elo | 1224 | 1539 |
| critpt | 2.6 | 15.1 |
| Fortress | 54.9 | 12.4 |
| gdpval | 8.5 | 43.7 |
| GPQA Diamond | 84.4 | 89.8 |
| HLE | 22.5 | 62.1 |
| multichallenge | 53.6 | 75.3 |
| OmniScience Accuracy | 39 | 52 |
| OmniScience Non-Hallucination | 12.6 | 50 |
| scicode | 43 | 58.2 |
| TauBench V3 - Banking | 9.7 | 31.8 |
| Terminal-Bench 2.1 | 28.5 | 80 |
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.