Model comparison
DeepSeek-R1 vs Qwen3 32B
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.2 on the Noometry Index.
Last verified . 22 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 22 benchmarks with published results for both. DeepSeek-R1 scores higher in 7 categories and Qwen3 32B in 2 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-R1 leads 46.3 to 37.7.
- The biggest single-benchmark swing is Aider Polyglot: 71.4% for DeepSeek-R1 and 40% for Qwen3 32B.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $0.70 / $2.80 for Qwen3 32B.
- DeepSeek-R1 accepts more context: 164K tokens versus 131K.
- Qwen3 32B has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | Qwen3 32B | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 42.3 | 39.2 |
| Released | 2025-01-20 | 2025-04 |
| Weights | Proprietary | Open |
| Context window | 164K | 131K |
| Max output | 64K | 16K |
| Input $ / M tokens | $0.50 | $0.70 |
| Output $ / M tokens | $2.15 | $2.80 |
| Results tracked | 52 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), Qwen3 32B: 37.7 (#190)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| Aider Polyglot | 71.4% | 40% |
| SciCode | 35.7% | 35.4% |
| LMArena Coding | 1427 | 1358 |
| WeirdML | 41.6% | — |
| LiveBench Coding | 66.7% | — |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use Qwen3 32B leads
DeepSeek-R1: 30.7 (#75), Qwen3 32B: 32.6 (#62)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 48.7% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning Qwen3 32B leads
DeepSeek-R1: 18.6 (#278), Qwen3 32B: 20.2 (#241)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| Kagi LLM Benchmark | 69.4% | 54.9% |
| CritPt | 1.1% | 0.3% |
| LMArena Hard Prompts | 1416 | 1334 |
| Epoch Capabilities Index | 141.29 | 138.51 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| ARC-AGI-1 | 21.2% | — |
| Chess Puzzles | — | 5% |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 67.5% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 17.3% |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), Qwen3 32B: 39.7 (#99)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 66.9% |
| LMArena Math | 1400 | 1399 |
| Omni-MATH | 42.4% | — |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), Qwen3 32B: 40.0 (#125)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| GPQA Diamond | 76.3% | 65.7% |
| Vectara Hallucination Rate | 11.3% | 5.9% |
| LMArena Expert | 1394 | 1362 |
| MMLU-Pro | 79.3% | — |
| Confabulations | 12.7% | — |
| GPQA (HELM) | 66.6% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), Qwen3 32B: 45.6 (#167)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| LMArena Non-English | 1412 | 1317 |
| LMArena Chinese | 1442 | 1357 |
| LMArena German | 1404 | 1341 |
| LMArena Russian | 1423 | 1311 |
| LMArena French | 1417 | — |
| LMArena Japanese | 1391 | — |
| LMArena Korean | 1360 | — |
| LMArena Spanish | 1411 | — |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), Qwen3 32B: 68.9 (#179)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| LMArena Instruction Following | 1382 | 1305 |
| LiveBench Instruction Following | 80.5% | — |
| IFEval | 78.4% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Qwen3 32B: 43.8 (#87)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| Fiction.LiveBench | 75% | 74.2% |
| LMArena Longer Query | 1391 | 1327 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), Qwen3 32B: 52.9 (#163)
| Benchmark | DeepSeek-R1 | Qwen3 32B |
|---|---|---|
| LMArena Text | 1428 | 1340 |
| LMArena Creative Writing | 1405 | 1297 |
| LMArena Multi-Turn | 1405 | 1331 |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
| WildBench | 82.8% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Qwen3 32B?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 39.2 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or Qwen3 32B?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Qwen3 32B lists at $0.70 and $2.80.
Is DeepSeek-R1 or Qwen3 32B better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 37.7 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-R1 and Qwen3 32B share?
22 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Qwen3 32B has 26.