| Benchmark | Muse Spark | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 28.7 | 39.7 |
| AA Intelligence | 44.3 | 53.2 |
| AA-LCR | 77 | 81.3 |
| AA-Omniscience | 7.2 | 28.1 |
| arena_elo | 1488 | 1491 |
| arena_text_factuality | 1467 | 1466 |
| Artificial Analysis Coding Index | 58.6 | 71.3 |
| coding_arena_elo | 1540 | 1539 |
| critpt | 11.3 | 15.1 |
| cybergym | 43.5 | 59 |
| deepsearchqa_f1 | 74.8 | 84.9 |
| Fortress | 20.2 | 12.4 |
| gdpval | 32.3 | 43.7 |
| GPQA Diamond | 89.5 | 89.8 |
| HLE | 58.4 | 62.1 |
| HLE Calibration | 50.3 | 23.4 |
| MCP Atlas | 82.2 | 88.1 |
| multichallenge | 75.5 | 75.3 |
| MultiNRC | 59 | 65.6 |
| OmniScience Accuracy | 49.6 | 52 |
| OmniScience Non-Hallucination | 26.8 | 50 |
| OR-Bench (FRR) | 8 | 4.8 |
| scicode | 51.5 | 58.2 |
| SWE-bench Pro | 55 | 61.5 |
| TauBench V3 - Banking | 19.6 | 31.8 |
| Terminal-Bench 2.1 | 62.2 | 80 |
| wmdp | 85.6 | 87 |
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.