# DeepSeek V4 Pro vs Qwen3 Max

> DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 43.7 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-pro-vs-qwen3-max
- Last updated: 2026-10-10
- Shared benchmarks: 30

## Summary

- They share 30 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 8 categories and Qwen3 Max in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 22.6.
- The biggest single-benchmark swing is NYT Connections (extended): 91.3% for DeepSeek V4 Pro and 30.1% for Qwen3 Max.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $1.20 / $6 for Qwen3 Max.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 262K.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 54.3 | 43.7 |
| Rank | 31 | 87 |
| Context | 1M | 262K |
| Input $/M | $0.66 | $1.20 |
| Output $/M | $1.98 | $6 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Pro: 52.4 (#34)
- Qwen3 Max: 43.0 (#93)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| LMArena Coding | 1470 | 1456 |
| ALE-Bench | 1,403 | 370.45 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| WeirdML | 66.2% | — |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.8 (#58)
- Qwen3 Max: —

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| Vending-Bench 2 | 3,285 | 71.56 |
| APEX-Agents | 47.3% | — |

## Reasoning

- DeepSeek V4 Pro: 56.5 (#24)
- Qwen3 Max: 22.6 (#190)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| Kagi LLM Benchmark | 53.5% | 72.5% |
| NYT Connections (extended) | 91.3% | 30.1% |
| Chess Puzzles | 47% | 4% |
| LMArena Hard Prompts | 1461 | 1448 |
| Mystery Game Puzzles | 43% | 5% |
| DTBench | 93.9% | 82.1% |
| LMCA | 45.5% | 28.3% |
| Epoch Capabilities Index | 155.31 | 142.38 |
| ARC-AGI-2 | 61.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |

## Math

- DeepSeek V4 Pro: 64.8 (#30)
- Qwen3 Max: 38.7 (#131)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 18.9% |
| OTIS Mock AIME 2024-2025 | 98.6% | 73.3% |
| LMArena Math | 1455 | 1446 |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| MATH Level 5 | — | 97.1% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#31)
- Qwen3 Max: 48.1 (#78)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| GPQA Diamond | 91.7% | 72.6% |
| SimpleQA Verified | 52.9% | 48.7% |
| LMArena Expert | 1464 | 1455 |
| Vectara Hallucination Rate | 8.6% | — |

## Multilingual

- DeepSeek V4 Pro: 54.4 (#45)
- Qwen3 Max: 53.7 (#62)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| LMArena Non-English | 1439 | 1429 |
| LMArena Chinese | 1486 | 1478 |
| LMArena French | 1472 | 1449 |
| LMArena German | 1458 | 1463 |
| LMArena Japanese | 1445 | 1397 |
| LMArena Korean | 1447 | 1399 |
| LMArena Russian | 1453 | 1428 |
| LMArena Spanish | 1458 | 1462 |

## Instruction Following

- DeepSeek V4 Pro: 76.1 (#47)
- Qwen3 Max: 74.8 (#87)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| LMArena Instruction Following | 1448 | 1419 |

## Long Context

- DeepSeek V4 Pro: 45.0 (#51)
- Qwen3 Max: 41.6 (#134)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| LMArena Longer Query | 1458 | 1438 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | — | 14.5% |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.5 (#46)
- Qwen3 Max: 62.4 (#76)

| Benchmark | DeepSeek V4 Pro | Qwen3 Max |
|---|---|---|
| LMArena Text | 1451 | 1439 |
| LMArena Creative Writing | 1446 | 1402 |
| LMArena Multi-Turn | 1467 | 1446 |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |

## FAQ

### Is DeepSeek V4 Pro better than Qwen3 Max?

DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 43.7 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Pro or Qwen3 Max?

DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Qwen3 Max lists at $1.20 and $6.

### Is DeepSeek V4 Pro or Qwen3 Max better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 43.0 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 Max share?

30 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Qwen3 Max has 33.