| Benchmark | Muse Spark | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 28.7 | 30.9 |
| AA Intelligence | 44.3 | 46.7 |
| AA-LCR | 77 | 74.7 |
| AA-Omniscience | 7.2 | 13.5 |
| arena_elo | 1488 | 1474 |
| arena_text_factuality | 1467 | 1479 |
| Artificial Analysis Coding Index | 58.6 | 66 |
| coding_arena_elo | 1540 | 1517 |
| critpt | 11.3 | 13.4 |
| gdpval | 32.3 | 38.5 |
| GPQA Diamond | 89.5 | 92.4 |
| HLE | 58.4 | 41.4 |
| IFBench | 75.9 | 80.5 |
| itbenchSre | 16.2 | 42.5 |
| MCP Atlas | 82.2 | 76.4 |
| OmniScience Accuracy | 49.6 | 31.1 |
| OmniScience Non-Hallucination | 26.8 | 74.4 |
| scicode | 51.5 | 53.5 |
| simpleqa_verified | 66.3 | 58.5 |
| SWE-bench Pro | 55 | 60.6 |
| SWE-bench Verified | 77.4 | 80.4 |
| TauBench V3 - Banking | 19.6 | 11.8 |
| Terminal-Bench 2.0 | 59 | 69.7 |
| Terminal-Bench 2.1 | 62.2 | 75 |
| Terminal-Bench Hard | 45.5 | 50.8 |
| τ²-Bench Telecom (AA run) | 91.5 | 94.7 |
| τ³-Bench | 19.6 | 10.9 |
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.