| Benchmark | Gemini 3.5 Flash | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 52 | 10 |
| AA-LCR | 81 | 20.3 |
| AA-Omniscience | 21.2 | -52.2 |
| Artificial Analysis Coding Index | 70.1 | 11.9 |
| critpt | 13.1 | 0 |
| GPQA Diamond | 92.2 | 49.1 |
| HLE | 42.7 | 4.2 |
| IFBench | 76.3 | 36.9 |
| livebench | 75 | 52.3 |
| OmniScience Accuracy | 51.4 | 17.6 |
| OmniScience Non-Hallucination | 38.2 | 15.3 |
| scicode | 53.1 | 26.7 |
| Terminal-Bench Hard | 46.2 | 4.5 |
| τ²-Bench Telecom (AA run) | 95.6 | 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.