# DeepSeek V4 Flash vs Qwen2-72B

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 29.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-qwen2-72b
- Last updated: 2026-10-10
- Shared benchmarks: 20

## Summary

- They share 20 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and Qwen2-72B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek V4 Flash leads 55.5 to 21.2.
- The biggest single-benchmark swing is WeirdML: 63% for DeepSeek V4 Flash and 11.3% for Qwen2-72B.

## Snapshot

| | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 53.6 | 29.5 |
| Rank | 32 | 294 |
| Context | 1M | — |
| Input $/M | $0.15 | — |
| Output $/M | $0.60 | — |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- Qwen2-72B: 28.7 (#303)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| WeirdML | 63% | 11.3% |
| LMArena Coding | 1457 | 1196 |
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1581 | — |
| SciCode | 49.9% | — |
| BigCodeBench Instruct | — | 38.5% |
| BigCodeBench Complete | — | 54% |
| ALE-Bench | 1,306 | — |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- Qwen2-72B: 16.9 (#138)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| TheAgentCompany | — | 1.1% |
| METR Time Horizons | — | 29.9% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- Qwen2-72B: 22.2 (#184)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Hard Prompts | 1443 | 1191 |
| Epoch Capabilities Index | 154.51 | 125.27 |
| ARC-AGI-2 | 61.4% | — |
| SimpleBench | 61.1% | — |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 89.6% | — |
| ARC-AGI-1 | 89% | — |
| CritPt | 16.6% | — |
| Chess Puzzles | 33% | — |
| Mystery Game Puzzles | 34% | — |
| DTBench | 90.9% | — |
| LMCA | 41.7% | — |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- Qwen2-72B: 27.9 (#232)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Math | 1427 | 1235 |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |
| OTIS Mock AIME 2024-2025 | 94.4% | — |
| ProofBench | 56% | — |
| MATH Level 5 | — | 39.1% |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- Qwen2-72B: 21.2 (#270)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| GPQA Diamond | 91% | 40.8% |
| LMArena Expert | 1441 | 1171 |
| SimpleQA Verified | 33.6% | — |
| MMLU | — | 82.4% |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- Qwen2-72B: 36.0 (#239)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Non-English | 1420 | 1176 |
| LMArena Chinese | 1467 | 1240 |
| LMArena French | 1438 | 1169 |
| LMArena German | 1418 | 1151 |
| LMArena Japanese | 1403 | 1111 |
| LMArena Korean | 1383 | 1083 |
| LMArena Russian | 1428 | 1169 |
| LMArena Spanish | 1437 | 1168 |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- Qwen2-72B: 62.0 (#236)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Instruction Following | 1421 | 1181 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- Qwen2-72B: 36.1 (#230)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Longer Query | 1435 | 1193 |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- Qwen2-72B: 41.0 (#236)

| Benchmark | DeepSeek V4 Flash | Qwen2-72B |
|---|---|---|
| LMArena Text | 1432 | 1203 |
| LMArena Creative Writing | 1403 | 1181 |
| LMArena Multi-Turn | 1449 | 1195 |
| EQ-Bench Creative Writing | 1559 | — |

## FAQ

### Is DeepSeek V4 Flash better than Qwen2-72B?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 29.5 on the Noometry Index.

### Is DeepSeek V4 Flash or Qwen2-72B better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 28.7 in the Noometry coding category.

### How many benchmarks do DeepSeek V4 Flash and Qwen2-72B share?

20 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Qwen2-72B has 26.