# DeepSeek V4 Flash vs Qwen3-30B-A3B

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 38.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-qwen3-30b-a3b
- Last updated: 2026-10-12
- Shared benchmarks: 28

## Summary

- They share 28 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and Qwen3-30B-A3B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 22.2.
- The biggest single-benchmark swing is WeirdML: 63% for DeepSeek V4 Flash and 29.8% for Qwen3-30B-A3B.
- Qwen3-30B-A3B is cheaper at $0.12 / $0.50 per million input/output tokens, against $0.15 / $0.60 for DeepSeek V4 Flash.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 41K.

## Snapshot

| | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| Provider | DeepSeek | Alibaba (Qwen) |
| Noometry Index | 53.6 | 38.9 |
| Rank | 35 | 179 |
| Context | 1M | 41K |
| Input $/M | $0.15 | $0.12 |
| Output $/M | $0.60 | $0.50 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Flash: 47.9 (#59)
- Qwen3-30B-A3B: 37.5 (#194)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| SciCode | 49.9% | 33.3% |
| WeirdML | 63% | 29.8% |
| LMArena Coding | 1457 | 1416 |
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1582 | — |
| ALE-Bench | 1,306 | — |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- Qwen3-30B-A3B: 29.8 (#82)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 41.4% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#30)
- Qwen3-30B-A3B: 22.2 (#204)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| Kagi LLM Benchmark | 52.2% | 54.9% |
| CritPt | 16.6% | 0.3% |
| Chess Puzzles | 33% | 8% |
| LMArena Hard Prompts | 1444 | 1398 |
| DTBench | 90.9% | 69.3% |
| LMCA | 41.7% | 22.4% |
| Epoch Capabilities Index | 154.49 | 139.63 |
| ARC-AGI-2 | 61.4% | — |
| SimpleBench | 61.1% | — |
| NYT Connections (extended) | 89.6% | — |
| ARC-AGI-1 | 89% | — |
| Mystery Game Puzzles | 34% | — |

## Math

- DeepSeek V4 Flash: 60.3 (#37)
- Qwen3-30B-A3B: 37.4 (#157)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| MathArena Final-Answer Competitions | 76.5% | 47.8% |
| OTIS Mock AIME 2024-2025 | 94.4% | 70.3% |
| LMArena Math | 1427 | 1394 |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| ProofBench | 56% | — |

## Knowledge

- DeepSeek V4 Flash: 55.4 (#48)
- Qwen3-30B-A3B: 41.8 (#105)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| GPQA Diamond | 91% | 70.1% |
| LMArena Expert | 1441 | 1396 |
| SimpleQA Verified | 33.6% | — |
| Confabulations | — | 12.3% |

## Multilingual

- DeepSeek V4 Flash: 53.0 (#72)
- Qwen3-30B-A3B: 49.5 (#132)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| LMArena Non-English | 1420 | 1372 |
| LMArena Chinese | 1468 | 1433 |
| LMArena French | 1439 | 1418 |
| LMArena German | 1418 | 1380 |
| LMArena Japanese | 1406 | 1337 |
| LMArena Korean | 1384 | 1331 |
| LMArena Russian | 1428 | 1370 |
| LMArena Spanish | 1436 | 1404 |

## Instruction Following

- DeepSeek V4 Flash: 74.9 (#81)
- Qwen3-30B-A3B: 72.0 (#142)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| LMArena Instruction Following | 1421 | 1363 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#85)
- Qwen3-30B-A3B: 31.0 (#283)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| LMArena Longer Query | 1434 | 1379 |
| Fiction.LiveBench | — | 40.6% |

## Writing & Preference

- DeepSeek V4 Flash: 63.8 (#61)
- Qwen3-30B-A3B: 55.6 (#143)

| Benchmark | DeepSeek V4 Flash | Qwen3-30B-A3B |
|---|---|---|
| LMArena Text | 1432 | 1384 |
| LMArena Creative Writing | 1403 | 1317 |
| LMArena Multi-Turn | 1449 | 1378 |
| Short-Story Creative Writing | — | 75.3% |
| EQ-Bench Creative Writing | 1559 | — |

## FAQ

### Is DeepSeek V4 Flash better than Qwen3-30B-A3B?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 38.9 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Flash or Qwen3-30B-A3B?

Qwen3-30B-A3B is cheaper. It lists at $0.12 per million input tokens and $0.50 per million output tokens; DeepSeek V4 Flash lists at $0.15 and $0.60.

### Is DeepSeek V4 Flash or Qwen3-30B-A3B better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 37.5 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 41K.

### How many benchmarks do DeepSeek V4 Flash and Qwen3-30B-A3B share?

28 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Qwen3-30B-A3B has 32.