| Benchmark | GPT-5.1 | Muse Spark |
|---|---|---|
| AA Agentic Index | 32.2 | 28.7 |
| AA Intelligence | 37.5 | 44.3 |
| AA-LCR | 76.7 | 77 |
| AA-Omniscience | 5.4 | 7.2 |
| ARC-AGI-2 | 17.6 | 42.5 |
| arena_elo | 1439 | 1488 |
| arena_text_factuality | 1445 | 1467 |
| arena_vision | 1250 | 1295 |
| Artificial Analysis Coding Index | 49.4 | 58.6 |
| coding_arena_elo | 1421 | 1540 |
| critpt | 4.9 | 11.3 |
| Fortress | 25.7 | 20.2 |
| frontiermath_tier_4 | 12.5 | 14.6 |
| gdpval | 25 | 32.3 |
| GPQA Diamond | 88.1 | 89.5 |
| HLE | 28.5 | 58.4 |
| IFBench | 72.9 | 75.9 |
| MCP Atlas | 50.1 | 82.2 |
| MMMU-Pro | 75.5 | 80.5 |
| multichallenge | 63.4 | 75.5 |
| MultiNRC | 49 | 59 |
| OmniScience Accuracy | 37.7 | 49.6 |
| OmniScience Non-Hallucination | 48.1 | 26.8 |
| scicode | 43.3 | 51.5 |
| simpleqa_verified | 48.9 | 66.3 |
| SWE-bench Verified | 76.3 | 77.4 |
| TauBench V3 - Banking | 15.9 | 19.6 |
| Terminal-Bench 2.0 | 47.6 | 59 |
| Terminal-Bench 2.1 | 52.4 | 62.2 |
| Terminal-Bench Hard | 45.5 | 45.5 |
| τ²-Bench Telecom (AA run) | 81.9 | 91.5 |
| τ³-Bench | 14 | 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.