DeepSeek · model
DeepSeek-V2.5 benchmark scores
As of 2026-10-29, DeepSeek-V2.5 (DeepSeek) has measured scores on 2 benchmarks and estimated scores on 21 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 42.3% | measured |
| AA-HLE | 4.3% | estimated ± 2.1 pp, high confidence |
| CritPt | 0.5% | estimated ± 3.0 pp, high confidence |
| GPQA Diamond | 87.0% | estimated ± 2.3 pp, low confidence |
| HLE | 5.1% | estimated ± 10.1 pp, low confidence |
| MMLU-Pro | 65.1% | estimated ± 4.1 pp, medium confidence |
| Artificial Analysis Intelligence Index | 6.6% | measured |
| GPQA | 47.3% | estimated ± 6.1 pp, medium confidence |
| MMLU | 79.9% | estimated ± 2.7 pp, medium confidence |
| AA-Omniscience Accuracy | 19.1% | estimated ± 7.2 pp, medium confidence |
| MMMLU | 73.6% | estimated ± 1.7 pp, medium confidence |
| AA MMLU-Pro | 83.2% | estimated ± 1.1 pp, medium confidence |
| ARC-AGI-1 | 70.0% | estimated ± 8.1 pp, low confidence |
| ARC-AGI-2 | 16.0% | estimated ± 14.9 pp, low confidence |
| Vals GPQA Diamond | 81.3% | estimated ± 5.3 pp, low confidence |
| Vals MMLU-Pro | 83.9% | estimated ± 2.1 pp, medium confidence |
| HLE w/o tools | 28.5% | estimated ± 4.6 pp, medium confidence |
| C-Eval | 87.6% | estimated ± 1.0 pp, low confidence |
| MMLU-Redux | 87.0% | estimated ± 1.2 pp, low confidence |
| SuperGPQA | 61.7% | estimated ± 10.1 pp, low confidence |
| HealthBench (length-adjusted) | 53.9% | estimated ± 3.3 pp, medium confidence |
| HealthBench Professional (raw) | 48.9% | estimated ± 8.6 pp, low confidence |
| SimpleQA | 9.6% | estimated ± 7.8 pp, low confidence |