| Benchmark | Claude Opus 4.6 | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 67.6 | 39.7 |
| AA Intelligence | 44.9 | 53.2 |
| AA-LCR | 74.3 | 81.3 |
| AA-Omniscience | 13.7 | 28.1 |
| arena_elo | 1497 | 1491 |
| arena_text_factuality | 1487 | 1466 |
| Artificial Analysis Coding Index | 48.1 | 71.3 |
| baby_vision | 14.8 | 76.3 |
| charxiv_reasoning_with_tools | 84.7 | 88.4 |
| coding_arena_elo | 1545 | 1539 |
| critpt | 12.6 | 15.1 |
| cybergym | 73.8 | 59 |
| deepsearchqa_f1 | 91.3 | 84.9 |
| Fortress | 20.5 | 12.4 |
| gdpval | 55.9 | 43.7 |
| GPQA Diamond | 91.3 | 89.8 |
| HLE | 53.1 | 62.1 |
| livebench_agentic_coding | 49 | 58.5 |
| livebench_coding | 78.2 | 77.2 |
| livebench_data_analysis | 69.9 | 72.5 |
| livebench_instruction_following | 63.3 | 69.6 |
| livebench_language | 83.3 | 74.3 |
| livebench_math | 89.3 | 87.1 |
| livebench_reasoning | 88.7 | 87.7 |
| MCP Atlas | 76.8 | 88.1 |
| multichallenge | 56 | 75.3 |
| MultiNRC | 57.1 | 65.6 |
| OmniScience Accuracy | 47 | 52 |
| OmniScience Non-Hallucination | 37.2 | 50 |
| OSWorld-Verified | 72.7 | 80.8 |
| scicode | 52 | 58.2 |
| SWE-bench Pro | 57.3 | 61.5 |
| toolathlon | 56.8 | 75.6 |
| VisualToolBench | 27.5 | 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.