| Benchmark | Gemini 3.6 Flash | GPT-5.6 Sol |
|---|---|---|
| AA Agentic Index | 40.5 | 57.8 |
| AA Intelligence | 52 | 61 |
| AA-LCR | 79 | 77.7 |
| AA-Omniscience | 22.1 | 22 |
| aime_2026 | 96.7 | 99.9 |
| arc_agi_1 | 34.5 | 97.5 |
| ARC-AGI-2 | 2.6 | 92.5 |
| arena_elo | 1481 | 1482 |
| arena_text_factuality | 1470 | 1474 |
| Artificial Analysis Coding Index | 69.2 | 78.3 |
| charxiv_rq | 85.2 | 87.8 |
| coding_arena_elo | 1538 | 1619 |
| critpt | 10.6 | 32.3 |
| DeepSWE | 49 | 73 |
| DeepSWE 1.1 | 49 | 73 |
| GDP (Surge AI) | 22 | 30.7 |
| gdpval | 46.1 | 61.1 |
| GDPval-AA v2 | 1421 | 1748 |
| GPQA Diamond | 92.8 | 94.6 |
| HLE | 40.8 | 49.5 |
| livebench_agentic_coding | 43.4 | 56.2 |
| livebench_coding | 77.9 | 83.9 |
| livebench_data_analysis | 63 | 79.8 |
| livebench_instruction_following | 75.4 | 71.8 |
| livebench_language | 83.9 | 87.7 |
| livebench_math | 86.4 | 96.2 |
| livebench_reasoning | 85.2 | 91.7 |
| MMMU-Pro | 83.2 | 84.6 |
| OmniScience Accuracy | 50 | 59.4 |
| OmniScience Non-Hallucination | 44.4 | 10.6 |
| OSWorld 2.0 | 33.8 | 62.6 |
| OSWorld-Verified | 83 | 83 |
| scicode | 52.7 | 56.9 |
| simpleqa_verified | 68.7 | 71.6 |
| SWE-bench Pro | 58.7 | 64.6 |
| TauBench V3 - Banking | 29.9 | 44.3 |
| Terminal-Bench 2.1 | 78 | 89.5 |
| Terminal-bench 3.0 | 5.4 | 34.6 |
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.