| Benchmark | Claude Opus 4.8 | Gemini 3.6 Flash |
|---|---|---|
| AA Agentic Index | 49.4 | 40.5 |
| AA Intelligence | 57.3 | 52 |
| AA-LCR | 73 | 79 |
| AA-Omniscience | 28.8 | 22.1 |
| aime_2026 | 100 | 96.7 |
| arc_agi_1 | 92.5 | 34.5 |
| ARC-AGI-2 | 72.1 | 2.6 |
| arena_elo | 1482 | 1481 |
| arena_text_factuality | 1464 | 1470 |
| Artificial Analysis Coding Index | 74.3 | 69.2 |
| charxiv_reasoning_with_tools | 89.9 | 89.4 |
| charxiv_rq | 80.5 | 85.2 |
| coding_arena_elo | 1539 | 1538 |
| critpt | 20.9 | 10.6 |
| DeepSWE | 59 | 49 |
| DeepSWE 1.1 | 59 | 49 |
| GDP (Surge AI) | 84.8 | 22 |
| gdpval | 54.2 | 46.1 |
| GDPval-AA v2 | 1593 | 1421 |
| GPQA Diamond | 93.6 | 92.8 |
| HLE | 57.9 | 40.8 |
| hmmt_feb_2026 | 96.7 | 89.4 |
| livebench_agentic_coding | 50.5 | 43.4 |
| livebench_coding | 81.8 | 77.9 |
| livebench_data_analysis | 66 | 63 |
| livebench_instruction_following | 72 | 75.4 |
| livebench_language | 79.7 | 83.9 |
| livebench_math | 94.3 | 86.4 |
| livebench_reasoning | 89.2 | 85.2 |
| matharena_visual_math_overall | 81.6 | 87.8 |
| MMMU-Pro | 78.9 | 83.2 |
| OmniScience Accuracy | 48.8 | 50 |
| OmniScience Non-Hallucination | 60.7 | 44.4 |
| OSWorld 2.0 | 55.7 | 33.8 |
| OSWorld-Verified | 83.4 | 83 |
| scicode | 53.5 | 52.7 |
| simpleqa_verified | 39.5 | 68.7 |
| SWE-bench Pro | 69.2 | 58.7 |
| TauBench V3 - Banking | 34.2 | 29.9 |
| Terminal-Bench 2.1 | 85 | 78 |
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.