| Benchmark | Claude 3.7 Sonnet | Muse Spark |
|---|---|---|
| AA Agentic Index | 37 | 28.7 |
| AA Intelligence | 27.6 | 44.3 |
| AA-LCR | 62.3 | 77 |
| AA-Omniscience | -0.7 | 7.2 |
| ARC-AGI-2 | 0 | 42.5 |
| arena_vision | 1195 | 1295 |
| Artificial Analysis Coding Index | 36.4 | 58.6 |
| critpt | 0.9 | 11.3 |
| Fortress | 38 | 20.2 |
| gdpval | 27.4 | 32.3 |
| GPQA Diamond | 84.8 | 89.5 |
| HLE | 9.7 | 58.4 |
| IFBench | 48.3 | 75.9 |
| MMMU-Pro | 60.1 | 80.5 |
| multichallenge | 51.6 | 75.5 |
| MultiNRC | 27.8 | 59 |
| OmniScience Accuracy | 28.2 | 49.6 |
| OmniScience Non-Hallucination | 60 | 26.8 |
| scicode | 40.3 | 51.5 |
| SWE-bench Verified | 70.3 | 77.4 |
| Terminal-Bench Hard | 21.2 | 45.5 |
| τ²-Bench Telecom (AA run) | 54.7 | 91.5 |
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.