DeepSeek
availableShows if the model has enough results for an index.DeepSeek V3
DeepSeek V3 is a non-reasoning model from DeepSeek in the DeepSeek family. 24 benchmarks count toward its score, in 6 categories.
IndexOverall score. 50 is the middle.32.6 ±4.8
CoverageShare of the index weight with results.85%
SpeedOutput tokens per second.—
Input / 1MUS dollars per 1M input tokens.$0.27
Output / 1MUS dollars per 1M output tokens.$1.1
ContextMaximum tokens in one request.128K
EloLMArena rating and rank.1332 (#209)
50 is the middle of the board. The range shows the doubt in the index.
21,770 votes. Elo shows what people prefer. It does not change the score.
CapabilitiesScore per category. 50 is the middle.
50 is the middleResults
24 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result on the index scale. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| MGSM | Multilingual | 91.9% | — | — | 9 Jan 2026 | Vals AI |
| MGSM | Multilingual | 91.7% | — | — | 9 Jan 2026 | Vals AI |
| MATH 500 | Math | 88.6% | 40.8 | — | 9 Jan 2026 | Vals AI |
| Instruction-Following Eval | Instruction | 86.1% | 36.8 | — | — | Jeffrey Zhou et al. |
| MATH 500 | Math | 84.5% | 36.2 | — | 9 Jan 2026 | Vals AI |
| MMLU Pro | Knowledge | 79.5% | 45.7 | — | 1 Sept 2026 | Vals AI |
| MMLU Pro | Knowledge | 76.6% | 41.3 | — | 1 Sept 2026 | Vals AI |
| Massive Multitask Language Understanding Professional | Knowledge | 75.9% | 40.1 | — | — | Yubo Wang et al. |
| LiveCodeBench | Coding | 65.5% | 43.9 | — | 1 Sept 2026 | Vals AI |
| LiveCodeBench | Coding | 65.5% | 43.9 | — | 1 Sept 2026 | Vals AI |
| MATH level 5 | Math | 64.9% | 30.5 | — | — | Epoch AI |
| GPQA Diamond | Knowledge | 61.6% | 35.1 | — | 1 Sept 2026 | Vals AI |
| Graduate-Level Google-Proof Q&A | Knowledge | 59.1% | 32.8 | — | — | David Rein et al. |
| GPQA Diamond | Knowledge | 58.1% | 31.9 | — | 1 Sept 2026 | Vals AI |
| GPQA diamond | Knowledge | 56.5% | 30.4 | — | — | Epoch AI |
| Artificial Analysis GPQA Diamond | Knowledge | 55.7% | 25.6 | — | — | Artificial Analysis |
| AIME | Math | 52.2% | 38.1 | — | 16 Apr 2026 | Vals AI |
| Software Engineering Benchmark Verified | Coding | 42.0% | 31.2 | — | — | Carlos E. Jimenez et al. |
| SWE-bench Verified | Coding | 42.0% | 31.2 | SWE-Exp | 1 Sept 2026 | SWE-bench team |
| AIME | Math | 39.9% | 32.3 | — | 16 Apr 2026 | Vals AI |
| LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code | Coding | 37.6% | 18.5 | — | — | Naman Jain et al. |
| SWE-bench Lite | Coding | 36.7% | 34.5 | KGCompass | 11 Sept 2025 | SWE-bench team |
| Artificial Analysis SciCode | Coding | 35.8% | 42.6 | — | — | Artificial Analysis |
| Artificial Analysis IFBench | Instruction | 34.8% | 24.4 | — | — | Artificial Analysis |
| SWE-bench Lite | Coding | 30.7% | 29.4 | Moatless Tools | 11 Sept 2025 | SWE-bench team |
| SWE-bench Lite | Coding | 30.7% | 29.4 | Moatless Tools | 11 Sept 2025 | SWE-bench team |
| Artificial Analysis Long Context Reasoning | Reasoning | 29.3% | 28.5 | — | — | Artificial Analysis |
| Artificial Analysis Omniscience Accuracy | Knowledge | 25.5% | 45.4 | — | — | Artificial Analysis |
| Artificial Analysis Coding Index | Coding | 23.0% | 35.2 | — | — | Artificial Analysis |
| τ²-Bench Tool-Agent-User Evaluation | Agentic | 22.8% | 15.2 | — | — | Victor Barres et al. |
| OTIS Mock AIME 2024-2025 | Math | 15.8% | 20.5 | — | — | Epoch AI |
| Artificial Analysis Intelligence Index | Knowledge | 8.5% | 33.0 | — | — | Artificial Analysis |
| Artificial Analysis Humanity's Last Exam | Knowledge | 2.9% | 27.7 | — | — | Artificial Analysis |
| FrontierMath-2025-02-28-Private | Math | 1.7% | 32.2 | — | — | Epoch AI |
| IOI v1 | Coding | 1.7% | 40.9 | — | 9 Aug 2026 | Vals AI |
| IOI v1 | Coding | 1.7% | 40.9 | — | 9 Aug 2026 | Vals AI |
| Artificial Analysis Agentic Index | Agentic | 0.8% | 37.8 | — | — | Artificial Analysis |
| GDPval-AA normalized | Agentic | 0.0% | 35.6 | — | — | Artificial Analysis |
| Critical Physics Tasks | Reasoning | 0.0% | 38.4 | — | — | Artificial Analysis |
24 benchmarks count, from 37 of 39 results. A grey row does not count. Too few models took that benchmark.