Model comparison
DeepSeek V4 Pro vs Qwen3-Coder 480B-A35B Instruct
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 38.1 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeek
Alibaba (Qwen)Summary
- They share 21 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 9 categories and Qwen3-Coder 480B-A35B Instruct in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 25.5.
- The biggest single-benchmark swing is WeirdML: 66.2% for DeepSeek V4 Pro and 41.2% for Qwen3-Coder 480B-A35B Instruct.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $1.50 / $7.50 for Qwen3-Coder 480B-A35B Instruct.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 262K.
Side by side
| DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct | |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 54.3 | 38.1 |
| Released | 2026-04-24 | 2025-04 |
| Weights | Open | Open |
| Context window | 1M | 262K |
| Max output | 393K | 66K |
| Input $ / M tokens | $0.66 | $1.50 |
| Output $ / M tokens | $1.98 | $7.50 |
| Results tracked | 48 | 25 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), Qwen3-Coder 480B-A35B Instruct: 35.5 (#223)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena WebDev | 1582 | 1275 |
| WeirdML | 66.2% | 41.2% |
| LMArena Coding | 1470 | 1412 |
| ALE-Bench | 1,403 | 461.45 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| SWE-bench Verified (bash only) | — | 55.4% |
| SciCode | 51% | — |
| GSO | — | 4.9% |
| AlgoTune | — | 1.44 |
Agentic & Tool Use DeepSeek V4 Pro leads
DeepSeek V4 Pro: 32.8 (#58), Qwen3-Coder 480B-A35B Instruct: 23.9 (#123)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| Terminal-Bench | — | 27.2% |
| APEX-Agents | 47.3% | — |
| Vending-Bench 2 | 3,285 | — |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), Qwen3-Coder 480B-A35B Instruct: 25.5 (#149)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| Kagi LLM Benchmark | 53.5% | 49.5% |
| LMArena Hard Prompts | 1461 | 1372 |
| ARC-AGI-2 | 61.3% | — |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| Chess Puzzles | 47% | — |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| Epoch Capabilities Index | 155.31 | — |
| ForecastBench | 56.1 | — |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), Qwen3-Coder 480B-A35B Instruct: 37.6 (#150)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Math | 1455 | 1365 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| OTIS Mock AIME 2024-2025 | 98.6% | — |
| ProofBench | 50% | — |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), Qwen3-Coder 480B-A35B Instruct: 37.0 (#162)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Expert | 1464 | 1338 |
| GPQA Diamond | 91.7% | — |
| SimpleQA Verified | 52.9% | — |
| Vectara Hallucination Rate | 8.6% | — |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), Qwen3-Coder 480B-A35B Instruct: 47.7 (#148)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Non-English | 1439 | 1346 |
| LMArena Chinese | 1486 | 1357 |
| LMArena French | 1472 | 1398 |
| LMArena German | 1458 | 1325 |
| LMArena Japanese | 1445 | 1310 |
| LMArena Korean | 1447 | 1305 |
| LMArena Russian | 1453 | 1366 |
| LMArena Spanish | 1458 | 1360 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), Qwen3-Coder 480B-A35B Instruct: 71.6 (#147)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Instruction Following | 1448 | 1355 |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), Qwen3-Coder 480B-A35B Instruct: 42.0 (#131)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Longer Query | 1458 | 1378 |
| CL-bench Life | 13.5% | — |
Writing & Preference DeepSeek V4 Pro leads
DeepSeek V4 Pro: 65.5 (#46), Qwen3-Coder 480B-A35B Instruct: 55.3 (#147)
| Benchmark | DeepSeek V4 Pro | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Text | 1451 | 1357 |
| LMArena Creative Writing | 1446 | 1333 |
| LMArena Multi-Turn | 1467 | 1365 |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than Qwen3-Coder 480B-A35B Instruct?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 38.1 on the Noometry Index.
Which is cheaper, DeepSeek V4 Pro or Qwen3-Coder 480B-A35B Instruct?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Qwen3-Coder 480B-A35B Instruct lists at $1.50 and $7.50.
Is DeepSeek V4 Pro or Qwen3-Coder 480B-A35B Instruct better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 35.5 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 262K.
How many benchmarks do DeepSeek V4 Pro and Qwen3-Coder 480B-A35B Instruct share?
21 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Qwen3-Coder 480B-A35B Instruct has 25.