| Benchmark | Claude Opus 4.5 | Gemini 3.6 Flash |
|---|---|---|
| AA Agentic Index | 59.6 | 40.5 |
| AA Intelligence | 41.9 | 52 |
| AA-LCR | 76 | 79 |
| AA-Omniscience | 14 | 22.1 |
| aime_2026 | 95.1 | 96.7 |
| arc_agi_1 | 82 | 34.5 |
| ARC-AGI-2 | 37.6 | 2.6 |
| arena_elo | 1473 | 1481 |
| arena_text_factuality | 1458 | 1470 |
| Artificial Analysis Coding Index | 47.8 | 69.2 |
| charxiv_rq | 68.5 | 85.2 |
| coding_arena_elo | 1495 | 1538 |
| critpt | 4.6 | 10.6 |
| gdpval | 47.3 | 46.1 |
| GPQA Diamond | 87 | 92.8 |
| HLE | 30.8 | 40.8 |
| hmmt_feb_2026 | 85.3 | 89.4 |
| livebench_agentic_coding | 39.7 | 43.4 |
| livebench_coding | 79.7 | 77.9 |
| livebench_data_analysis | 74.4 | 63 |
| livebench_instruction_following | 62.5 | 75.4 |
| livebench_language | 81.3 | 83.9 |
| livebench_math | 90.4 | 86.4 |
| livebench_reasoning | 80.1 | 85.2 |
| MMMU-Pro | 74 | 83.2 |
| OmniScience Accuracy | 46.6 | 50 |
| OmniScience Non-Hallucination | 39 | 44.4 |
| OSWorld-Verified | 66.3 | 83 |
| scicode | 50 | 52.7 |
| simpleqa_verified | 41.8 | 68.7 |
| SWE-bench Pro | 57.1 | 58.7 |
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.