| Benchmark | Gemini 3 Pro | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 52 | 39.7 |
| AA Intelligence | 40.6 | 53.2 |
| AA-LCR | 73 | 81.3 |
| AA-Omniscience | 15.3 | 28.1 |
| arena_elo | 1485 | 1491 |
| arena_text_factuality | 1481 | 1466 |
| Artificial Analysis Coding Index | 46.5 | 71.3 |
| coding_arena_elo | 1438 | 1539 |
| critpt | 9.1 | 15.1 |
| cybergym | 39.9 | 59 |
| deepsearchqa_f1 | 63.2 | 84.9 |
| Fortress | 41.7 | 12.4 |
| gdpval | 34.2 | 43.7 |
| GPQA Diamond | 91.9 | 89.8 |
| HLE | 45.8 | 62.1 |
| MCP Atlas | 70.3 | 88.1 |
| multichallenge | 65.7 | 75.3 |
| MultiNRC | 59 | 65.6 |
| OmniScience Accuracy | 55.8 | 52 |
| OmniScience Non-Hallucination | 10 | 50 |
| scicode | 56.1 | 58.2 |
| SWE-bench Pro | 43.3 | 61.5 |
| toolathlon | 36.4 | 75.6 |
| VisualToolBench | 26.9 | 44.8 |
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.