| Benchmark | Gemini 3 Flash Preview | Qwen3.5 397B A17B |
|---|---|---|
| AA Agentic Index | 49.7 | 53.3 |
| AA Intelligence | 38.7 | 34.3 |
| AA-LCR | 73 | 72.7 |
| AA-Omniscience | 10.1 | -30.8 |
| aime_2026 | 96.7 | 94.2 |
| apexAgents | 27.7 | 15.3 |
| arena_elo | 1473 | 1442 |
| arena_text_factuality | 1452 | 1445 |
| arena_vision | 1260 | 1249 |
| Artificial Analysis Coding Index | 42.6 | 48.2 |
| charxiv_rq | 80.3 | 80.8 |
| coding_arena_elo | 1387 | 1399 |
| critpt | 8.6 | 1.7 |
| gdpval | 35.2 | 35.8 |
| GPQA Diamond | 90.4 | 89.3 |
| HLE | 43.5 | 29 |
| hmmt_feb_2025 | 97.5 | 94.8 |
| hmmt_feb_2026 | 89.4 | 87.9 |
| hmmt_nov_2025 | 93.3 | 92.7 |
| IFBench | 78 | 78.8 |
| mmmlu | 91.8 | 88.5 |
| MMMU-Pro | 81.2 | 79 |
| OmniScience Accuracy | 53.4 | 30.8 |
| OmniScience Non-Hallucination | 7.6 | 17.3 |
| scicode | 50.6 | 42 |
| simpleqa_verified | 67.4 | 26 |
| SWE-bench Multilingual | 72.7 | 69.3 |
| SWE-bench Pro | 49.6 | 50.9 |
| SWE-bench Verified | 78 | 76.4 |
| TauBench V3 - Banking | 20.8 | 13.4 |
| Terminal-Bench 2.0 | 47.6 | 52.5 |
| Terminal-Bench 2.1 | 58 | 51.3 |
| Terminal-Bench Hard | 38.6 | 40.9 |
| toolathlon | 49.4 | 38.3 |
| VideoMMMU | 86.9 | 84.7 |
| τ²-Bench Telecom (AA run) | 80.4 | 95.6 |
| τ³-Bench | 17.5 | 13.4 |
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.