# DeepSeek-V3 vs Qwen3 Max

> Qwen3 Max is the stronger model overall, scoring 43.7 to 39.5 on the Noometry Index. DeepSeek-V3 costs 5.9× less per token, which makes it the better buy when Qwen3 Max's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-v3-vs-qwen3-max
- Last updated: 2026-10-12
- Shared benchmarks: 25

## Summary

- They share 25 benchmarks with published results for both. DeepSeek-V3 scores higher in 0 categories and Qwen3 Max in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen3 Max leads 48.1 to 37.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 73.3% for Qwen3 Max.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $1.20 / $6 for Qwen3 Max.
- Qwen3 Max accepts more context: 262K tokens versus 164K.
- DeepSeek-V3 has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 39.5 | 43.7 |
| Rank | 166 | 87 |
| Context | 164K | 262K |
| Input $/M | $0.24 | $1.20 |
| Output $/M | $0.90 | $6 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek-V3: 42.3 (#106)
- Qwen3 Max: 43.0 (#93)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| LMArena Coding | 1368 | 1456 |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| ALE-Bench | — | 370.45 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |

## Agentic & Tool Use

- DeepSeek-V3: —
- Qwen3 Max: —

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| METR Time Horizons | 49.6% | — |
| Vending-Bench 2 | — | 71.56 |

## Reasoning

- DeepSeek-V3: 20.5 (#236)
- Qwen3 Max: 22.6 (#190)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| Kagi LLM Benchmark | 52.3% | 72.5% |
| LMArena Hard Prompts | 1365 | 1448 |
| DTBench | 64.8% | 82.1% |
| LMCA | 15.5% | 28.3% |
| Epoch Capabilities Index | 135.94 | 142.38 |
| SimpleBench | 27.2% | — |
| NYT Connections (extended) | — | 30.1% |
| CritPt | 0% | — |
| Chess Puzzles | — | 4% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 5% |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |

## Math

- DeepSeek-V3: 32.1 (#219)
- Qwen3 Max: 38.7 (#131)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 73.3% |
| LMArena Math | 1373 | 1446 |
| MATH Level 5 | 75.5% | 97.1% |
| FrontierMath (Tiers 1-3) | — | 18.9% |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |

## Knowledge

- DeepSeek-V3: 37.5 (#155)
- Qwen3 Max: 48.1 (#78)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| GPQA Diamond | 67.6% | 72.6% |
| LMArena Expert | 1351 | 1455 |
| SimpleQA Verified | — | 48.7% |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |

## Multilingual

- DeepSeek-V3: 48.5 (#143)
- Qwen3 Max: 53.7 (#62)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| LMArena Non-English | 1358 | 1429 |
| LMArena Chinese | 1391 | 1478 |
| LMArena French | 1385 | 1449 |
| LMArena German | 1374 | 1463 |
| LMArena Japanese | 1333 | 1397 |
| LMArena Korean | 1319 | 1399 |
| LMArena Russian | 1373 | 1428 |
| LMArena Spanish | 1358 | 1462 |

## Instruction Following

- DeepSeek-V3: 72.8 (#130)
- Qwen3 Max: 74.8 (#87)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| LMArena Instruction Following | 1345 | 1419 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |

## Long Context

- DeepSeek-V3: 34.0 (#253)
- Qwen3 Max: 41.6 (#134)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| Fiction.LiveBench | 50% | 66.7% |
| LMArena Longer Query | 1352 | 1438 |
| CL-bench | — | 14.5% |

## Writing & Preference

- DeepSeek-V3: 57.4 (#130)
- Qwen3 Max: 62.4 (#76)

| Benchmark | DeepSeek-V3 | Qwen3 Max |
|---|---|---|
| LMArena Text | 1375 | 1439 |
| LMArena Creative Writing | 1364 | 1402 |
| LMArena Multi-Turn | 1389 | 1446 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |

## FAQ

### Is DeepSeek-V3 better than Qwen3 Max?

Qwen3 Max is the stronger model overall, scoring 43.7 to 39.5 on the Noometry Index. DeepSeek-V3 costs 5.9× less per token, which makes it the better buy when Qwen3 Max's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek-V3 or Qwen3 Max?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Qwen3 Max lists at $1.20 and $6.

### Is DeepSeek-V3 or Qwen3 Max better for coding?

They score almost the same on coding (42.3 vs 43.0); test both on your own repository before choosing.

### Which has the bigger context window?

Qwen3 Max does, with 262K tokens against 164K.

### How many benchmarks do DeepSeek-V3 and Qwen3 Max share?

25 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Qwen3 Max has 33.