| Benchmark | DeepSeek-R1 | o1 |
|---|---|---|
| AA Agentic Index | 3.1 | 31.1 |
| AA Intelligence | 18.6 | 24 |
| AA-LCR | 56 | 63.3 |
| AA-Omniscience | -31.3 | -10.6 |
| aider_polyglot | 71.6 | 61.7 |
| AIME 2024 | 91.4 | 79.2 |
| AIME 2025 | 87.5 | 81.7 |
| AIR-Bench 2024 | 52.9 | 80 |
| anthropic_red_team | 99.1 | 98.3 |
| Artificial Analysis Coding Index | 24.6 | 39.7 |
| bbq | 96.6 | 97.3 |
| Codeforces | 2029 | 2061 |
| Codeforces (Percentile) | 96.3 | 96.6 |
| Codeforces (Rating) | 2029 | 2061 |
| critpt | 0.6 | 0.3 |
| DROP (3-shot F1) | 92.2 | 90.2 |
| Fortress | 74.4 | 19.4 |
| gdpval | 1.5 | 11.5 |
| GPQA Diamond | 81 | 75.7 |
| harmbench | 54.6 | 96.3 |
| HLE | 17.7 | 7.7 |
| IFBench | 39 | 70.3 |
| livecodebench_pass1cot | 65.9 | 63.4 |
| MATH | 97.3 | 96.4 |
| math_500_em | 98 | 96.4 |
| mmlu | 92.9 | 91.8 |
| OmniScience Accuracy | 30.7 | 34.7 |
| OmniScience Non-Hallucination | 10.5 | 30.7 |
| scicode | 35.7 | 35.8 |
| simple_safety_tests | 98.3 | 99 |
| simplebench | 40.8 | 41.7 |
| simpleqa | 92.3 | 47 |
| SWE-bench Verified | 57.6 | 48.9 |
| TAU-bench (airline) | 53.5 | 50 |
| TAU-bench (retail) | 63.9 | 70.8 |
| Terminal-Bench Hard | 6.1 | 12.9 |
| xstest | 98.8 | 97 |
| τ²-Bench Telecom (AA run) | 11.4 | 62.6 |
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.