Model comparison
DeepSeek-V3 vs QwQ-32B
DeepSeek-V3 and QwQ-32B score almost the same on the Noometry Index (39.5 vs 39.8), so choose on price, context window or the category you care about most.
Last verified . 35 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 35 benchmarks with published results for both. DeepSeek-V3 scores higher in 5 categories and QwQ-32B in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in long context, where QwQ-32B leads 49.0 to 34.0.
- The biggest single-benchmark swing is Aider Polyglot: 55.1% for DeepSeek-V3 and 20.9% for QwQ-32B.
Side by side
| DeepSeek-V3 | QwQ-32B | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 39.5 | 39.8 |
| Released | 2024-12-26 | 2024-11-28 |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), QwQ-32B: 35.4 (#226)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| Aider Polyglot | 55.1% | 20.9% |
| BigCodeBench Instruct | 50% | 44.6% |
| LiveBench Coding | 70.9% | 72.2% |
| LMArena Coding | 1368 | 1333 |
| BigCodeBench Complete | 62.2% | 54.4% |
| SciCode | 35.8% | — |
| WeirdML | 36.1% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, QwQ-32B: —
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning QwQ-32B leads
DeepSeek-V3: 20.5 (#236), QwQ-32B: 23.7 (#174)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| LiveBench Reasoning | 65.8% | 83.5% |
| LMArena Hard Prompts | 1365 | 1325 |
| LiveBench Data Analysis | 60.9% | 65% |
| Epoch Capabilities Index | 135.94 | 137.6 |
| ForecastBench | 59.1 | 58.3 |
| LiveBench | 66.9% | 72% |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| Chess Puzzles | — | 5% |
| DTBench | 64.8% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math QwQ-32B leads
DeepSeek-V3: 32.1 (#219), QwQ-32B: 38.0 (#143)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 59.2% |
| LiveBench Math | 73.5% | 77.8% |
| LMArena Math | 1373 | 1359 |
| Omni-MATH | 40.3% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge Too close to call
DeepSeek-V3: 37.5 (#155), QwQ-32B: 37.2 (#158)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| GPQA Diamond | 67.6% | 65.3% |
| Confabulations | 26.1% | 15.6% |
| LMArena Expert | 1351 | 1324 |
| MMLU-Pro | 72.3% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), QwQ-32B: 44.8 (#176)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| LMArena Non-English | 1358 | 1305 |
| LMArena Chinese | 1391 | 1378 |
| LMArena French | 1385 | 1336 |
| LMArena German | 1374 | 1313 |
| LMArena Japanese | 1333 | 1262 |
| LMArena Korean | 1319 | 1279 |
| LMArena Russian | 1373 | 1297 |
| LMArena Spanish | 1358 | 1354 |
Instruction Following Too close to call
DeepSeek-V3: 72.8 (#130), QwQ-32B: 72.6 (#137)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 81.8% |
| LMArena Instruction Following | 1345 | 1297 |
| IFEval | 83.2% | — |
Long Context QwQ-32B leads
DeepSeek-V3: 34.0 (#253), QwQ-32B: 49.0 (#11)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| Fiction.LiveBench | 50% | 83.3% |
| LMArena Longer Query | 1352 | 1308 |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), QwQ-32B: 50.6 (#180)
| Benchmark | DeepSeek-V3 | QwQ-32B |
|---|---|---|
| LMArena Text | 1375 | 1329 |
| LMArena Creative Writing | 1364 | 1288 |
| Short-Story Creative Writing | 77% | 80.2% |
| EQ-Bench Creative Writing | 1472 | 1257 |
| LMArena Multi-Turn | 1389 | 1314 |
| LiveBench Language | 49.1% | 51.4% |
| WildBench | 83% | — |
Frequently asked questions
Is DeepSeek-V3 better than QwQ-32B?
DeepSeek-V3 and QwQ-32B score almost the same on the Noometry Index (39.5 vs 39.8), so choose on price, context window or the category you care about most.
Is DeepSeek-V3 or QwQ-32B better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 35.4 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and QwQ-32B share?
35 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and QwQ-32B has 36.