# DeepSeek V4 Flash vs Qwen2.5 7B Instruct

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 28.6 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-qwen2-5-7b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 6

## Summary

- They share 6 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 6 categories and Qwen2.5 7B Instruct in 0 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Flash leads 60.0 to 11.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 94.4% for DeepSeek V4 Flash and 2.5% for Qwen2.5 7B Instruct.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.17 / $0.70 for Qwen2.5 7B Instruct.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 131K.

## Snapshot

| | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 53.6 | 28.6 |
| Rank | 32 | 314 |
| Context | 1M | 131K |
| Input $/M | $0.15 | $0.17 |
| Output $/M | $0.60 | $0.70 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- Qwen2.5 7B Instruct: 35.9 (#204)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1581 | — |
| SciCode | 49.9% | — |
| WeirdML | 63% | — |
| BigCodeBench Instruct | — | 37.6% |
| LMArena Coding | 1457 | — |
| BigCodeBench Complete | — | 46.1% |
| ALE-Bench | 1,306 | — |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- Qwen2.5 7B Instruct: 23.5 (#117)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| BALROG | — | 7.8% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- Qwen2.5 7B Instruct: 14.3 (#317)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| Chess Puzzles | 33% | 0% |
| DTBench | 90.9% | 47.7% |
| LMCA | 41.7% | 6.4% |
| Epoch Capabilities Index | 154.51 | 118.44 |
| 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% | — |
| LMArena Hard Prompts | 1443 | — |
| Mystery Game Puzzles | 34% | — |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- Qwen2.5 7B Instruct: 11.8 (#300)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 94.4% | 2.5% |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |
| ProofBench | 56% | — |
| Omni-MATH | — | 29.4% |
| LMArena Math | 1427 | — |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- Qwen2.5 7B Instruct: 17.0 (#281)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| GPQA Diamond | 91% | 35.5% |
| SimpleQA Verified | 33.6% | — |
| MMLU-Pro | — | 53.9% |
| GPQA (HELM) | — | 34.1% |
| LMArena Expert | 1441 | — |
| MMLU | — | 72.9% |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- Qwen2.5 7B Instruct: —

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Non-English | 1420 | — |
| LMArena Chinese | 1467 | — |
| LMArena French | 1438 | — |
| LMArena German | 1418 | — |
| LMArena Japanese | 1403 | — |
| LMArena Korean | 1383 | — |
| LMArena Russian | 1428 | — |
| LMArena Spanish | 1437 | — |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- Qwen2.5 7B Instruct: 63.4 (#226)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| IFEval | — | 74.1% |
| LMArena Instruction Following | 1421 | — |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- Qwen2.5 7B Instruct: —

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Longer Query | 1435 | — |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- Qwen2.5 7B Instruct: 48.8 (#194)

| Benchmark | DeepSeek V4 Flash | Qwen2.5 7B Instruct |
|---|---|---|
| LMArena Text | 1432 | — |
| LMArena Creative Writing | 1403 | — |
| EQ-Bench Creative Writing | 1559 | — |
| WildBench | — | 73.1% |
| LMArena Multi-Turn | 1449 | — |

## FAQ

### Is DeepSeek V4 Flash better than Qwen2.5 7B Instruct?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 28.6 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Flash or Qwen2.5 7B Instruct?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Qwen2.5 7B Instruct lists at $0.17 and $0.70.

### Is DeepSeek V4 Flash or Qwen2.5 7B Instruct better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 35.9 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 131K.

### How many benchmarks do DeepSeek V4 Flash and Qwen2.5 7B Instruct share?

6 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Qwen2.5 7B Instruct has 15.