| Benchmark | GPT-5 | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 49.7 | 39.7 |
| AA Intelligence | 35.3 | 53.2 |
| AA-LCR | 76.3 | 81.3 |
| AA-Omniscience | -8.7 | 28.1 |
| Artificial Analysis Coding Index | 38.9 | 71.3 |
| coding_arena_elo | 1419 | 1539 |
| critpt | 5.7 | 15.1 |
| Fortress | 17 | 12.4 |
| gdpval | 32.5 | 43.7 |
| GPQA Diamond | 87.3 | 89.8 |
| HLE | 28.5 | 62.1 |
| multichallenge | 71.1 | 75.3 |
| MultiNRC | 52.1 | 65.6 |
| OmniScience Accuracy | 40.3 | 52 |
| OmniScience Non-Hallucination | 21 | 50 |
| scicode | 43 | 58.2 |
| SWE-bench Pro | 41.8 | 61.5 |
| TauBench V3 - Banking | 22.1 | 31.8 |
| Terminal-Bench 2.1 | 35.2 | 80 |
| VisualToolBench | 18.7 | 44.8 |
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.