Model family
DeepSeek DeepSeek
DeepSeek chat and reasoning models, including R1-style reasoners.
Published scores mapped to this family
| Source | Benchmark | Published id | Score | Rank | As of |
|---|---|---|---|---|---|
| Hugging Face | MMLU-Pro | deepseek-ai/DeepSeek-V4-Pro | 87.5 | 5 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | MMLU-Pro | deepseek-ai/DeepSeek-V4-Flash | 86.4 | 10 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | MMLU-Pro | deepseek-ai/DeepSeek-R1-0528 | 85 | 16 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | MMLU-Pro | deepseek-ai/DeepSeek-R1 | 84 | 19 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | MMLU-Pro | deepseek-ai/DeepSeek-V3-0324 | 81.2 | 33 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | Humanity's Last Exam | deepseek-ai/DeepSeek-V4.1-Flash | 63.9 | 1 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | Humanity's Last Exam | deepseek-ai/DeepSeek-V4-Pro | 37.7 | 28 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | Humanity's Last Exam | deepseek-ai/DeepSeek-V4.1-Flash | 36.8 | 31 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | Humanity's Last Exam | deepseek-ai/DeepSeek-V4-Flash | 34.8 | 33 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | GPQA | deepseek-ai/DeepSeek-V4.1-Flash | 90.9 | 10 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | GPQA | deepseek-ai/DeepSeek-V4-Pro | 90.1 | 13 | Oct 7, 2026, 8:57 AM UTC |
| Hugging Face | GPQA | deepseek-ai/DeepSeek-V4-Flash | 88.1 | 22 | Oct 7, 2026, 8:57 AM UTC |
| HELM | HELM Capabilities | deepseek-ai_deepseek-r1-0528 | 69.9 | 24 | Oct 7, 2026, 8:57 AM UTC |
| HELM | HELM Capabilities | deepseek-ai_deepseek-v3 | 66.5 | 31 | Oct 7, 2026, 8:57 AM UTC |