| Benchmark | DeepSeek-V4-Pro | Gemini 1.5 Pro |
|---|---|---|
| AA Intelligence | 53 | 10 |
| Artificial Analysis Coding Index | 59.4 | 23.6 |
| Chinese SimpleQA (C-SimpleQA) | 77.7 | 59.4 |
| GPQA Diamond | 90.5 | 59.1 |
| HLE | 48.2 | 4.9 |
| LiveCodeBench | 93.5 | 30.5 |
| MMLU-Pro | 87.5 | 75.8 |
| scicode | 50 | 29.5 |
| simplebench | 50.9 | 27.1 |
| simpleqa | 57.9 | 23.4 |
| SWE-bench Verified | 80.6 | 34.2 |
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.