# DeepSeek V4 Flash vs Qwen1.5-72B

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 30.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-qwen1-5-72b
- Last updated: 2026-10-10
- Shared benchmarks: 18

## Summary

- They share 18 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and Qwen1.5-72B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek V4 Flash leads 55.5 to 11.5.
- The biggest single-benchmark swing is GPQA Diamond: 91% for DeepSeek V4 Flash and 28.8% for Qwen1.5-72B.

## Snapshot

| | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 53.6 | 30.3 |
| Rank | 32 | 282 |
| Context | 1M | — |
| Input $/M | $0.15 | — |
| Output $/M | $0.60 | — |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- Qwen1.5-72B: 31.4 (#272)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Coding | 1457 | 1166 |
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1581 | — |
| SciCode | 49.9% | — |
| WeirdML | 63% | — |
| BigCodeBench Instruct | — | 33.2% |
| BigCodeBench Complete | — | 40.3% |
| ALE-Bench | 1,306 | — |
| HumanEval+ | — | 59.1% |
| MBPP+ | — | 61.6% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- Qwen1.5-72B: 21.2 (#201)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Hard Prompts | 1443 | 1148 |
| 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% | — |
| Epoch Capabilities Index | 154.51 | — |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- Qwen1.5-72B: 30.7 (#205)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Math | 1427 | 1164 |
| 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% | — |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- Qwen1.5-72B: 11.5 (#295)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| GPQA Diamond | 91% | 28.8% |
| LMArena Expert | 1441 | 1136 |
| SimpleQA Verified | 33.6% | — |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- Qwen1.5-72B: 33.4 (#248)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Non-English | 1420 | 1135 |
| LMArena Chinese | 1467 | 1186 |
| LMArena French | 1438 | 1158 |
| LMArena German | 1418 | 1084 |
| LMArena Japanese | 1403 | 1061 |
| LMArena Korean | 1383 | 1049 |
| LMArena Russian | 1428 | 1104 |
| LMArena Spanish | 1437 | 1110 |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- Qwen1.5-72B: 59.6 (#251)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Instruction Following | 1421 | 1142 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- Qwen1.5-72B: 35.0 (#238)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Longer Query | 1435 | 1157 |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- Qwen1.5-72B: 37.4 (#253)

| Benchmark | DeepSeek V4 Flash | Qwen1.5-72B |
|---|---|---|
| LMArena Text | 1432 | 1166 |
| LMArena Creative Writing | 1403 | 1137 |
| LMArena Multi-Turn | 1449 | 1160 |
| EQ-Bench Creative Writing | 1559 | — |

## FAQ

### Is DeepSeek V4 Flash better than Qwen1.5-72B?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 30.3 on the Noometry Index.

### Is DeepSeek V4 Flash or Qwen1.5-72B better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 31.4 in the Noometry coding category.

### How many benchmarks do DeepSeek V4 Flash and Qwen1.5-72B share?

18 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Qwen1.5-72B has 22.