| Benchmark | Claude Opus 4.8 | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 49.4 | 39.7 |
| AA Intelligence | 57.3 | 53.2 |
| AA-LCR | 73 | 81.3 |
| AA-Omniscience | 28.8 | 28.1 |
| AIRS-Bench | 84 | 77 |
| arena_elo | 1482 | 1491 |
| arena_text_factuality | 1464 | 1466 |
| Artificial Analysis Coding Index | 74.3 | 71.3 |
| charxiv_reasoning_with_tools | 89.9 | 88.4 |
| coding_arena_elo | 1539 | 1539 |
| critpt | 20.9 | 15.1 |
| cybergym | 83.1 | 59 |
| deepsearchqa_f1 | 93.1 | 84.9 |
| DeepSWE 1.1 | 59 | 53.3 |
| Finance Agent v2 | 53.9 | 57.2 |
| Fortress | 18.2 | 12.4 |
| gdpval | 54.2 | 43.7 |
| GPQA Diamond | 93.6 | 89.8 |
| healthbench_professional | 57.4 | 59.3 |
| HLE | 57.9 | 62.1 |
| HLE Calibration | 26.5 | 23.4 |
| JobBench | 48.4 | 54.7 |
| livebench_agentic_coding | 50.5 | 58.5 |
| livebench_coding | 81.8 | 77.2 |
| livebench_data_analysis | 66 | 72.5 |
| livebench_instruction_following | 72 | 69.6 |
| livebench_language | 79.7 | 74.3 |
| livebench_math | 94.3 | 87.1 |
| livebench_reasoning | 89.2 | 87.7 |
| MCP Atlas | 83.6 | 88.1 |
| OmniScience Accuracy | 48.8 | 52 |
| OmniScience Non-Hallucination | 60.7 | 50 |
| OR-Bench (FRR) | 3.3 | 4.8 |
| OSWorld-Verified | 83.4 | 80.8 |
| scicode | 53.5 | 58.2 |
| SWE-bench Pro | 69.2 | 61.5 |
| TauBench V3 - Banking | 34.2 | 31.8 |
| Terminal-Bench 2.1 | 85 | 80 |
| toolathlon | 59.9 | 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.