| Benchmark | Claude Opus 4.5 | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 59.6 | 39.7 |
| AA Intelligence | 41.9 | 53.2 |
| AA-LCR | 76 | 81.3 |
| AA-Omniscience | 14 | 28.1 |
| arena_elo | 1473 | 1491 |
| arena_text_factuality | 1458 | 1466 |
| Artificial Analysis Coding Index | 47.8 | 71.3 |
| coding_arena_elo | 1495 | 1539 |
| critpt | 4.6 | 15.1 |
| cybergym | 50.6 | 59 |
| deepsearchqa_f1 | 76.1 | 84.9 |
| Fortress | 13.6 | 12.4 |
| gdpval | 47.3 | 43.7 |
| GPQA Diamond | 87 | 89.8 |
| HLE | 30.8 | 62.1 |
| livebench_agentic_coding | 39.7 | 58.5 |
| livebench_coding | 79.7 | 77.2 |
| livebench_data_analysis | 74.4 | 72.5 |
| livebench_instruction_following | 62.5 | 69.6 |
| livebench_language | 81.3 | 74.3 |
| livebench_math | 90.4 | 87.1 |
| livebench_reasoning | 80.1 | 87.7 |
| MCP Atlas | 69.8 | 88.1 |
| multichallenge | 59 | 75.3 |
| MultiNRC | 48.6 | 65.6 |
| OmniScience Accuracy | 46.6 | 52 |
| OmniScience Non-Hallucination | 39 | 50 |
| OSWorld-Verified | 66.3 | 80.8 |
| scicode | 50 | 58.2 |
| SWE-bench Pro | 57.1 | 61.5 |
| toolathlon | 43.5 | 75.6 |
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.