Model comparison
DeepSeek-V3 vs Qwen3 235B-A22B
Qwen3 235B-A22B is the stronger model overall, scoring 43.2 to 39.2 on the Noometry Index. DeepSeek-V3 costs 3.0× less per token, which makes it the better buy when Qwen3 235B-A22B's lead doesn't matter for your workload.
Last verified . 41 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 41 benchmarks with published results for both. DeepSeek-V3 scores higher in 2 categories and Qwen3 235B-A22B in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 235B-A22B leads 48.5 to 30.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 86.7% for Qwen3 235B-A22B.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $0.70 / $2.80 for Qwen3 235B-A22B.
- DeepSeek-V3 accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3 | Qwen3 235B-A22B | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 39.2 | 43.2 |
| Released | 2024-12-26 | 2025-04 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 164K | 16K |
| Input $ / M tokens | $0.24 | $0.70 |
| Output $ / M tokens | $0.90 | $2.80 |
| Results tracked | 60 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 235B-A22B leads
DeepSeek-V3: 41.9 (#99), Qwen3 235B-A22B: 43.8 (#72)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| Aider Polyglot | 55.1% | 59.6% |
| SciCode | 35.8% | 42.4% |
| WeirdML | 36.1% | 41% |
| LMArena Coding | 1368 | 1445 |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Qwen3 235B-A22B: 33.7 (#48)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 52.1% |
| METR Time Horizons | 49.6% | — |
| Vending-Bench 2 | — | -11.34 |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.0 (#226), Qwen3 235B-A22B: 15.8 (#302)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| SimpleBench | 27.2% | 31% |
| Kagi LLM Benchmark | 52.3% | 69.4% |
| CritPt | 0% | 0% |
| LMArena Hard Prompts | 1365 | 1433 |
| DTBench | 64.8% | 80.3% |
| LMCA | 15.5% | 29.3% |
| Epoch Capabilities Index | 135.95 | 143.87 |
| ForecastBench | 59.1 | 59.7 |
| ARC-AGI-2 | — | 1.3% |
| ARC-AGI-1 | — | 11% |
| Chess Puzzles | — | 12% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 9% |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Qwen3 235B-A22B leads
DeepSeek-V3: 30.4 (#208), Qwen3 235B-A22B: 48.5 (#51)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 86.7% |
| Omni-MATH | 40.3% | 71.8% |
| LMArena Math | 1373 | 1431 |
| MATH Level 5 | 75.5% | 68.9% |
| FrontierMath (Feb 2025 set) | 1.7% | 8.5% |
| LiveBench Math | 73.5% | — |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Qwen3 235B-A22B leads
DeepSeek-V3: 37.5 (#147), Qwen3 235B-A22B: 49.5 (#67)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| GPQA Diamond | 67.6% | 80.1% |
| MMLU-Pro | 72.3% | 84.4% |
| Confabulations | 26.1% | 15.6% |
| Vectara Hallucination Rate | 6.1% | 9.3% |
| GPQA (HELM) | 53.8% | 72.7% |
| LMArena Expert | 1352 | 1460 |
| SimpleQA Verified | — | 40.4% |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual Qwen3 235B-A22B leads
DeepSeek-V3: 48.8 (#138), Qwen3 235B-A22B: 52.5 (#84)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| LMArena Non-English | 1358 | 1410 |
| LMArena Chinese | 1390 | 1481 |
| LMArena French | 1386 | 1445 |
| LMArena German | 1374 | 1432 |
| LMArena Japanese | 1333 | 1400 |
| LMArena Korean | 1321 | 1391 |
| LMArena Russian | 1372 | 1411 |
| LMArena Spanish | 1358 | 1429 |
Instruction Following Too close to call
DeepSeek-V3: 73.0 (#125), Qwen3 235B-A22B: 72.8 (#132)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| IFEval | 83.2% | 83.5% |
| LMArena Instruction Following | 1345 | 1408 |
| LiveBench Instruction Following | 81.5% | — |
Long Context Qwen3 235B-A22B leads
DeepSeek-V3: 34.0 (#248), Qwen3 235B-A22B: 46.1 (#25)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| Fiction.LiveBench | 50% | 75% |
| LMArena Longer Query | 1353 | 1426 |
Writing & Preference Qwen3 235B-A22B leads
DeepSeek-V3: 57.6 (#124), Qwen3 235B-A22B: 59.8 (#103)
| Benchmark | DeepSeek-V3 | Qwen3 235B-A22B |
|---|---|---|
| LMArena Text | 1375 | 1419 |
| LMArena Creative Writing | 1364 | 1384 |
| Short-Story Creative Writing | 77% | 83% |
| EQ-Bench Creative Writing | 1472 | 1366 |
| WildBench | 83% | 86.6% |
| LMArena Multi-Turn | 1389 | 1432 |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Qwen3 235B-A22B?
Qwen3 235B-A22B is the stronger model overall, scoring 43.2 to 39.2 on the Noometry Index. DeepSeek-V3 costs 3.0× less per token, which makes it the better buy when Qwen3 235B-A22B's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3 or Qwen3 235B-A22B?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Qwen3 235B-A22B lists at $0.70 and $2.80.
Is DeepSeek-V3 or Qwen3 235B-A22B better for coding?
Qwen3 235B-A22B scores higher on coding benchmarks: 43.8 versus 41.9 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3 and Qwen3 235B-A22B share?
41 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Qwen3 235B-A22B has 49.