| Benchmark | Claude Opus 4.8 | DeepSeek-V3 |
|---|---|---|
| AA Agentic Index | 49.4 | 1.6 |
| AA Intelligence | 57.3 | 15.4 |
| AA-LCR | 73 | 41.3 |
| AA-Omniscience | 28.8 | -37.6 |
| Artificial Analysis Coding Index | 74.3 | 23 |
| critpt | 20.9 | 0 |
| gdpval | 54.2 | 0 |
| GPQA Diamond | 93.6 | 68.4 |
| HLE | 57.9 | 5.2 |
| IFBench | 62.2 | 41 |
| livebench | 77.2 | 72.4 |
| longbench_v2 | 69.1 | 48.7 |
| OmniScience Accuracy | 48.8 | 25.4 |
| OmniScience Non-Hallucination | 60.7 | 23.3 |
| scicode | 53.5 | 35.8 |
| simplebench | 64.8 | 27.2 |
| SWE-bench Verified | 88.6 | 42 |
| TauBench V3 - Banking | 34.2 | 4.7 |
| Terminal-Bench 2.1 | 85 | 16.9 |
| Terminal-Bench Hard | 58.3 | 15.2 |
| τ²-Bench Telecom (AA run) | 94.4 | 47.1 |
| τ³-Bench | 27.6 | 4.7 |
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.