Model comparison
DeepSeek-V3 vs Qwen2-72B
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 30.0 on the Noometry Index.
Last verified . 25 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 25 benchmarks with published results for both. DeepSeek-V3 scores higher in 6 categories and Qwen2-72B in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.4 to 40.8.
- The biggest single-benchmark swing is MATH Level 5: 75.5% for DeepSeek-V3 and 39.1% for Qwen2-72B.
Side by side
| DeepSeek-V3 | Qwen2-72B | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 39.5 | 30.0 |
| Released | 2024-12-26 | 2024-06-07 |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Qwen2-72B: 29.1 (#310)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| WeirdML | 36.1% | 11.3% |
| BigCodeBench Instruct | 50% | 38.5% |
| LMArena Coding | 1368 | 1196 |
| BigCodeBench Complete | 62.2% | 54% |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| LiveBench Coding | 70.9% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Qwen2-72B: 17.0 (#146)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| METR Time Horizons | 49.6% | 29.9% |
| TheAgentCompany | — | 1.1% |
Reasoning Qwen2-72B leads
DeepSeek-V3: 20.5 (#236), Qwen2-72B: 23.2 (#181)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Hard Prompts | 1365 | 1191 |
| Epoch Capabilities Index | 135.94 | 125.28 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 65.8% | — |
| DTBench | 64.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Qwen2-72B: 30.2 (#236)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Math | 1373 | 1235 |
| MATH Level 5 | 75.5% | 39.1% |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Qwen2-72B: 21.2 (#275)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| GPQA Diamond | 67.6% | 40.8% |
| LMArena Expert | 1351 | 1171 |
| MMLU | 87.2% | 82.4% |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Qwen2-72B: 35.9 (#244)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Non-English | 1358 | 1176 |
| LMArena Chinese | 1391 | 1240 |
| LMArena French | 1385 | 1170 |
| LMArena German | 1374 | 1151 |
| LMArena Japanese | 1333 | 1111 |
| LMArena Korean | 1319 | 1083 |
| LMArena Russian | 1373 | 1169 |
| LMArena Spanish | 1358 | 1169 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Qwen2-72B: 61.7 (#241)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Instruction Following | 1345 | 1181 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context Qwen2-72B leads
DeepSeek-V3: 34.0 (#253), Qwen2-72B: 36.1 (#235)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Longer Query | 1352 | 1192 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Qwen2-72B: 40.8 (#241)
| Benchmark | DeepSeek-V3 | Qwen2-72B |
|---|---|---|
| LMArena Text | 1375 | 1203 |
| LMArena Creative Writing | 1364 | 1181 |
| LMArena Multi-Turn | 1389 | 1196 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Qwen2-72B?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 30.0 on the Noometry Index.
Is DeepSeek-V3 or Qwen2-72B better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 29.1 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Qwen2-72B share?
25 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Qwen2-72B has 26.