| Benchmark | Claude Sonnet 4.6 | Muse Spark |
|---|---|---|
| AA Agentic Index | 61.6 | 28.7 |
| AA Intelligence | 48.4 | 44.3 |
| AA-LCR | 74 | 77 |
| AA-Omniscience | 12.2 | 7.2 |
| ARC-AGI-2 | 60.4 | 42.5 |
| arena_elo | 1472 | 1488 |
| arena_text_factuality | 1460 | 1467 |
| arena_vision | 1278 | 1295 |
| Artificial Analysis Coding Index | 63 | 58.6 |
| coding_arena_elo | 1523 | 1540 |
| critpt | 3.1 | 11.3 |
| frontiermath_tier_4 | 8.3 | 14.6 |
| gdpval | 54.8 | 32.3 |
| GPQA Diamond | 89.9 | 89.5 |
| HLE | 49 | 58.4 |
| IFBench | 56.6 | 75.9 |
| itbenchSre | 39.8 | 16.2 |
| MCP Atlas | 69.5 | 82.2 |
| MMMU-Pro | 75.6 | 80.5 |
| OmniScience Accuracy | 40.9 | 49.6 |
| OmniScience Non-Hallucination | 51.6 | 26.8 |
| scicode | 47 | 51.5 |
| simpleqa_verified | 29 | 66.3 |
| SWE-bench Pro | 58.1 | 55 |
| SWE-bench Verified | 79.6 | 77.4 |
| TauBench V3 - Banking | 34.4 | 19.6 |
| Terminal-Bench 2.0 | 59.1 | 59 |
| Terminal-Bench 2.1 | 71.2 | 62.2 |
| Terminal-Bench Hard | 59.1 | 45.5 |
| τ²-Bench Telecom (AA run) | 97.9 | 91.5 |
| τ³-Bench | 30.5 | 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.