| Benchmark | Muse Spark | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 44.3 | 10 |
| AA-LCR | 77 | 20.3 |
| AA-Omniscience | 7.2 | -52.2 |
| Artificial Analysis Coding Index | 58.6 | 11.9 |
| critpt | 11.3 | 0 |
| Fortress | 20.2 | 56.4 |
| GPQA Diamond | 89.5 | 49.1 |
| HLE | 58.4 | 4.2 |
| IFBench | 75.9 | 36.9 |
| OmniScience Accuracy | 49.6 | 17.6 |
| OmniScience Non-Hallucination | 26.8 | 15.3 |
| scicode | 51.5 | 26.7 |
| SWE-bench Verified | 77.4 | 23.8 |
| Terminal-Bench Hard | 45.5 | 4.5 |
| τ²-Bench Telecom (AA run) | 91.5 | 34.5 |
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.