| Benchmark | GPT-5.5 | Muse Spark 1.1 |
|---|---|---|
| AA Agentic Index | 47.4 | 39.7 |
| AA Intelligence | 56.3 | 53.2 |
| AA-LCR | 79 | 81.3 |
| AA-Omniscience | 20.5 | 28.1 |
| AIRS-Bench | 86 | 77 |
| arena_elo | 1482 | 1491 |
| arena_text_factuality | 1482 | 1466 |
| Artificial Analysis Coding Index | 74.9 | 71.3 |
| coding_arena_elo | 1457 | 1539 |
| critpt | 27.1 | 15.1 |
| cybergym | 81.8 | 59 |
| DeepSWE 1.1 | 67 | 53.3 |
| Finance Agent v2 | 51.8 | 57.2 |
| Fortress | 16.3 | 12.4 |
| gdpval | 49.5 | 43.7 |
| GPQA Diamond | 93.6 | 89.8 |
| healthbench_professional | 51.8 | 59.3 |
| HLE | 52.2 | 62.1 |
| HLE Calibration | 44.2 | 23.4 |
| JobBench | 38.3 | 54.7 |
| livebench_agentic_coding | 54 | 58.5 |
| livebench_coding | 82.2 | 77.2 |
| livebench_data_analysis | 81.6 | 72.5 |
| livebench_instruction_following | 70.7 | 69.6 |
| livebench_language | 87.4 | 74.3 |
| livebench_math | 95.9 | 87.1 |
| livebench_reasoning | 89.7 | 87.7 |
| MCP Atlas | 82.8 | 88.1 |
| OmniScience Accuracy | 58 | 52 |
| OmniScience Non-Hallucination | 12 | 50 |
| OR-Bench (FRR) | 6.7 | 4.8 |
| OSWorld-Verified | 78.7 | 80.8 |
| scicode | 56.1 | 58.2 |
| SWE-bench Pro | 58.6 | 61.5 |
| TauBench V3 - Banking | 39 | 31.8 |
| Terminal-Bench 2.1 | 84.3 | 80 |
| toolathlon | 55.6 | 75.6 |
| wmdp | 85.5 | 87 |
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.