Model comparison
DeepSeek-R1 vs QwQ-32B
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.8 on the Noometry Index.
Last verified . 33 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 33 benchmarks with published results for both. DeepSeek-R1 scores higher in 5 categories and QwQ-32B in 3 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-R1 leads 46.3 to 35.4.
- The biggest single-benchmark swing is Aider Polyglot: 71.4% for DeepSeek-R1 and 20.9% for QwQ-32B.
- QwQ-32B has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | QwQ-32B | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 42.3 | 39.8 |
| Released | 2025-01-20 | 2024-11-28 |
| Weights | Proprietary | Open |
| Context window | 164K | — |
| Max output | 64K | — |
| Input $ / M tokens | $0.50 | — |
| Output $ / M tokens | $2.15 | — |
| Results tracked | 52 | 36 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), QwQ-32B: 35.4 (#226)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| Aider Polyglot | 71.4% | 20.9% |
| LiveBench Coding | 66.7% | 72.2% |
| LMArena Coding | 1427 | 1333 |
| SciCode | 35.7% | — |
| WeirdML | 41.6% | — |
| BigCodeBench Instruct | — | 44.6% |
| BigCodeBench Complete | — | 54.4% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use Not comparable
DeepSeek-R1: 30.7 (#75), QwQ-32B: —
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning QwQ-32B leads
DeepSeek-R1: 18.6 (#278), QwQ-32B: 23.7 (#174)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| LiveBench Reasoning | 83.2% | 83.5% |
| LMArena Hard Prompts | 1416 | 1325 |
| LiveBench Data Analysis | 69.8% | 65% |
| Epoch Capabilities Index | 141.29 | 137.6 |
| ForecastBench | 60 | 58.3 |
| LiveBench | 71.6% | 72% |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | — | 5% |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), QwQ-32B: 38.0 (#143)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 59.2% |
| LiveBench Math | 80.7% | 77.8% |
| LMArena Math | 1400 | 1359 |
| Omni-MATH | 42.4% | — |
| MATH Level 5 | 96.6% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), QwQ-32B: 37.2 (#158)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| GPQA Diamond | 76.3% | 65.3% |
| Confabulations | 12.7% | 15.6% |
| LMArena Expert | 1394 | 1324 |
| MMLU-Pro | 79.3% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), QwQ-32B: 44.8 (#176)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| LMArena Non-English | 1412 | 1305 |
| LMArena Chinese | 1442 | 1378 |
| LMArena French | 1417 | 1336 |
| LMArena German | 1404 | 1313 |
| LMArena Japanese | 1391 | 1262 |
| LMArena Korean | 1360 | 1279 |
| LMArena Russian | 1423 | 1297 |
| LMArena Spanish | 1411 | 1354 |
Instruction Following Too close to call
DeepSeek-R1: 72.0 (#143), QwQ-32B: 72.6 (#137)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| LiveBench Instruction Following | 80.5% | 81.8% |
| LMArena Instruction Following | 1382 | 1297 |
| IFEval | 78.4% | — |
Long Context QwQ-32B leads
DeepSeek-R1: 45.4 (#36), QwQ-32B: 49.0 (#11)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| Fiction.LiveBench | 75% | 83.3% |
| LMArena Longer Query | 1391 | 1308 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), QwQ-32B: 50.6 (#180)
| Benchmark | DeepSeek-R1 | QwQ-32B |
|---|---|---|
| LMArena Text | 1428 | 1329 |
| LMArena Creative Writing | 1405 | 1288 |
| Short-Story Creative Writing | 83% | 80.2% |
| EQ-Bench Creative Writing | 1500 | 1257 |
| LMArena Multi-Turn | 1405 | 1314 |
| LiveBench Language | 48.5% | 51.4% |
| WildBench | 82.8% | — |
Frequently asked questions
Is DeepSeek-R1 better than QwQ-32B?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.8 on the Noometry Index.
Is DeepSeek-R1 or QwQ-32B better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 35.4 in the Noometry coding category.
How many benchmarks do DeepSeek-R1 and QwQ-32B share?
33 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and QwQ-32B has 36.