DeepSeek-V3.2-Exp ranks first on 2 of the 43 benchmarks it has scores on.
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 | 97.1% | #1 of 59 | Model card | |
| Vectara Hallucination Leaderboard (2025 dataset) | 5.30% | #15 of 104 | Official leaderboard | |
| MMLU-Pro | 85% | #27 of 151 | Model card | |
| Fiction.LiveBench | 83.3% | #11 of 57 | Epoch AI | |
| LMArena (Elo) | 1,425 | #97 of 333 | Official leaderboard | |
| Vectara Hallucination Leaderboard (HHEM) | 4.60% | #70 of 148 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 60.2 | #62 of 119 | Official leaderboard | |
| Slop Score | 54.2 | #9 of 17 | Official leaderboard | |
| Humanity's Last Exam | 19.8% | #99 of 147 | Model card | |
| CL-bench Life | 9.50% | #11 of 14 | Epoch AI | |
| CL-bench | 13.2% | #18 of 20 | Epoch AI |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BRUMO 2025 | 95.8% | #12 of 42 | Official leaderboard | |
| DTBench | 79.5% | #60 of 178 | Official leaderboard | |
| AIME 2025 | 89.3% | #53 of 143 | Model card | |
| GPQA Diamond | 79.9% | #112 of 284 | Model card | |
| CritPt | 2.86% | #80 of 160 | Epoch AI | |
| SMT 2025 | 84.9% | #21 of 40 | Official leaderboard | |
| CMIMC 2025 | 76.3% | #21 of 33 | Official leaderboard | |
| HMMT 2025 | 83.6% | #33 of 49 | Model card | |
| HMMT Nov 2025 | 84.2% | #18 of 22 | Official leaderboard | |
| MathArena (Apex, IMO, USAMO) | 0.52% | #44 of 47 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Aider Polyglot | 74.2% | #7 of 54 | Epoch AI | |
| LiveCodeBench | 74.1% | #21 of 73 | Model card | |
| Codeforces rating | 2,121 | #17 of 41 | Model card | |
| SWE-bench Verified | 67.8% | #91 of 150 | Model card | |
| WeirdML | 39.5% | #79 of 112 | Official leaderboard | |
| WebDev Arena | 1,367 | #77 of 105 | Epoch AI | |
| SWE-bench Multilingual | 57.9% | #59 of 70 | Model card | |
| Code Arena | 1,271 | #108 of 124 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BFCL V4 | 56.7% | #11 of 65 | Official leaderboard | |
| Step Race | 4 | #10 of 59 | Official leaderboard | |
| Terminal-Bench | 37.7% | #13 of 22 | Model card | |
| Toolathlon | 20.1% | #38 of 50 | Official leaderboard | |
| MCP-Universe | 19.9% | #25 of 30 | Official leaderboard | |
| BrowseComp | 40.1% | #67 of 77 | Model card | |
| BrowseComp-ZH | 47.9% | #16 of 17 | Model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| TheAgentCompany | 42.9% | #1 of 14 | Epoch AI | |
| LEXam | 57.4% | #7 of 31 | Official leaderboard | |
| ProfBench | 48.2% | #34 of 72 | Official leaderboard | |
| Vending-Bench 2 | $1,034 | #52 of 65 | Epoch AI |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Conceptual Reasoning Index | 45.7 | #82 of 149 | Official leaderboard | |
| ACCoRD | 61.8 | #83 of 150 | Official leaderboard | |
| LMCA | 29.1% | #86 of 139 | Epoch AI |
DeepSeek's models in color, against every other model on the Epoch Capabilities Index, by release date.
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it.