| Benchmark | Gemini 3.6 Flash | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 40.5 | 30.9 |
| AA Intelligence | 52 | 46.7 |
| AA-LCR | 79 | 74.7 |
| AA-Omniscience | 22.1 | 13.5 |
| aime_2026 | 96.7 | 97 |
| arena_elo | 1481 | 1474 |
| arena_text_factuality | 1470 | 1479 |
| Artificial Analysis Coding Index | 69.2 | 66 |
| coding_arena_elo | 1538 | 1517 |
| critpt | 10.6 | 13.4 |
| DeepSWE | 49 | 18 |
| DeepSWE 1.1 | 49 | 21.6 |
| gdpval | 46.1 | 38.5 |
| GPQA Diamond | 92.8 | 92.4 |
| HLE | 40.8 | 41.4 |
| hmmt_feb_2026 | 89.4 | 97.1 |
| livebench_agentic_coding | 43.4 | 43.6 |
| livebench_coding | 77.9 | 74.2 |
| livebench_data_analysis | 63 | 71.8 |
| livebench_instruction_following | 75.4 | 74 |
| livebench_language | 83.9 | 79.7 |
| livebench_math | 86.4 | 85.3 |
| livebench_reasoning | 85.2 | 83.3 |
| OmniScience Accuracy | 50 | 31.1 |
| OmniScience Non-Hallucination | 44.4 | 74.4 |
| scicode | 52.7 | 53.5 |
| simpleqa_verified | 68.7 | 58.5 |
| SWE-bench Pro | 58.7 | 60.6 |
| TauBench V3 - Banking | 29.9 | 11.8 |
| Terminal-Bench 2.1 | 78 | 75 |
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.