DeepSeek-V3.1-Terminus places in the top three on SimpleQA.
DeepSeek Sep 2025 Open weights Flagship
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SimpleQA | 96.8% | #2 of 59 | Model card | |
| MMLU-Pro | 85% | #26 of 151 | Model card | |
| LMArena (Elo) | 1,416 | #109 of 333 | Official leaderboard | |
| Humanity's Last Exam | 21.7% | #95 of 147 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 80.7% | #108 of 284 | Model card | |
| DTBench | 68.9% | #83 of 178 | Official leaderboard | |
| CritPt | 1.71% | #85 of 160 | Epoch AI |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Aider Polyglot | 76.1% | #6 of 29 | Model card | |
| LiveCodeBench | 74.9% | #18 of 73 | Model card | |
| StructEval | 77.2% | #7 of 28 | Official leaderboard | |
| SWE-bench Verified | 68.4% | #86 of 150 | Model card | |
| Codeforces rating | 2,046 | #24 of 41 | Model card | |
| ALE-Bench | 745 | #78 of 121 | Epoch AI | |
| SWE-bench Multilingual | 57.8% | #60 of 70 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Meta Agents Research Environments (ARE) / Gaia2 | 23.1% | #4 of 13 | Official leaderboard | |
| MCPMark | 21.3% | #15 of 28 | Official leaderboard | |
| SnakeBench | 21 | #117 of 183 | Official leaderboard | |
| CompileBench | 73.3% | #14 of 20 | Official leaderboard | |
| Terminal-Bench | 36.7% | #16 of 22 | Model card | |
| MCP-Universe | 21.6% | #22 of 30 | Official leaderboard | |
| BrowseComp | 38.5% | #69 of 77 | Model card | |
| BrowseComp-ZH | 45% | #17 of 17 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Conceptual Reasoning Index | 42.0 | #93 of 149 | Official leaderboard | |
| LMCA | 28.6% | #89 of 139 | Epoch AI | |
| ACCoRD | 55.2 | #105 of 150 | Official leaderboard |
The benchmarks the lab chose to headline, with the scores it gave.