| Benchmark | Muse Spark 1.1 | Qwen3.6 27B |
|---|---|---|
| AA Agentic Index | 39.7 | 27.5 |
| AA Intelligence | 53.2 | 37.7 |
| AA-LCR | 81.3 | 73.3 |
| AA-Omniscience | 28.1 | -20 |
| Artificial Analysis Coding Index | 71.3 | 53.7 |
| baby_vision | 76.3 | 28.9 |
| critpt | 15.1 | 1.1 |
| DeepSWE 1.1 | 53.3 | 13.3 |
| gdpval | 43.7 | 32 |
| GPQA Diamond | 89.8 | 87.8 |
| HLE | 62.1 | 24 |
| JobBench | 54.7 | 21.8 |
| livebench_agentic_coding | 58.5 | 39.3 |
| livebench_coding | 77.2 | 71.8 |
| livebench_data_analysis | 72.5 | 70.4 |
| livebench_instruction_following | 69.6 | 53.2 |
| livebench_language | 74.3 | 63.3 |
| livebench_math | 87.1 | 79.9 |
| livebench_reasoning | 87.7 | 70.3 |
| OmniScience Accuracy | 52 | 19.6 |
| OmniScience Non-Hallucination | 50 | 50.7 |
| OSWorld-Verified | 80.8 | 63.9 |
| scicode | 58.2 | 39.8 |
| SWE-bench Pro | 61.5 | 53.5 |
| TauBench V3 - Banking | 31.8 | 16.7 |
| Terminal-Bench 2.1 | 80 | 60.7 |
| WebArena-Verified | 69 | 48.8 |
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.