| Benchmark | Gemini 3.6 Flash | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 52 | 10 |
| AA-LCR | 79 | 20.3 |
| AA-Omniscience | 22.1 | -52.2 |
| Artificial Analysis Coding Index | 69.2 | 11.9 |
| critpt | 10.6 | 0 |
| GPQA Diamond | 92.8 | 49.1 |
| HLE | 40.8 | 4.2 |
| OmniScience Accuracy | 50 | 17.6 |
| OmniScience Non-Hallucination | 44.4 | 15.3 |
| scicode | 52.7 | 26.7 |
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.