# DeepSeek V4 Flash vs GPT-4 Turbo

> 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-gpt-4-turbo
- Last updated: 2026-10-10
- Shared benchmarks: 26

## Summary

- They share 26 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and GPT-4 Turbo in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Flash leads 60.0 to 8.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 94.4% for DeepSeek V4 Flash and 6.7% for GPT-4 Turbo.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 128K.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 53.6 | 30.3 |
| Rank | 32 | 281 |
| Context | 1M | 128K |
| Input $/M | $0.15 | $10 |
| Output $/M | $0.60 | $30 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Flash: 47.6 (#56)
- GPT-4 Turbo: 33.4 (#247)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| WeirdML | 63% | 18% |
| LMArena Coding | 1457 | 1268 |
| FrontierCode | 18.8% | — |
| LMArena WebDev | 1581 | — |
| SciCode | 49.9% | — |
| BigCodeBench Instruct | — | 48.2% |
| BigCodeBench Complete | — | 58.2% |
| ALE-Bench | 1,306 | — |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73.3% |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- GPT-4 Turbo: —

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| METR Time Horizons | — | 36.7% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#25)
- GPT-4 Turbo: 14.9 (#311)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| SimpleBench | 61.1% | 25.1% |
| Chess Puzzles | 33% | 6% |
| LMArena Hard Prompts | 1443 | 1251 |
| DTBench | 90.9% | 61.6% |
| LMCA | 41.7% | 9.8% |
| Epoch Capabilities Index | 154.51 | 127.25 |
| ARC-AGI-2 | 61.4% | — |
| Kagi LLM Benchmark | 52.2% | — |
| NYT Connections (extended) | 89.6% | — |
| ARC-AGI-1 | 89% | — |
| CritPt | 16.6% | — |
| Mystery Game Puzzles | 34% | — |
| ForecastBench | — | 59.4 |

## Math

- DeepSeek V4 Flash: 60.0 (#31)
- GPT-4 Turbo: 8.6 (#316)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 0.7% |
| OTIS Mock AIME 2024-2025 | 94.4% | 6.7% |
| LMArena Math | 1427 | 1272 |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |
| ProofBench | 56% | — |
| MATH Level 5 | — | 46.7% |

## Knowledge

- DeepSeek V4 Flash: 55.5 (#43)
- GPT-4 Turbo: 24.3 (#263)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| GPQA Diamond | 91% | 46.6% |
| LMArena Expert | 1441 | 1223 |
| SimpleQA Verified | 33.6% | — |
| Confabulations | — | 28.4% |
| MMLU | — | 81.3% |

## Multimodal

- DeepSeek V4 Flash: —
- GPT-4 Turbo: 30.0 (#104)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| LMArena Vision | — | 1090 |

## Multilingual

- DeepSeek V4 Flash: 53.2 (#71)
- GPT-4 Turbo: 40.7 (#211)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| LMArena Non-English | 1420 | 1245 |
| LMArena Chinese | 1467 | 1242 |
| LMArena French | 1438 | 1275 |
| LMArena German | 1418 | 1259 |
| LMArena Japanese | 1403 | 1194 |
| LMArena Korean | 1383 | 1187 |
| LMArena Russian | 1428 | 1259 |
| LMArena Spanish | 1437 | 1260 |

## Instruction Following

- DeepSeek V4 Flash: 75.1 (#76)
- GPT-4 Turbo: 66.0 (#211)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| LMArena Instruction Following | 1421 | 1250 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#81)
- GPT-4 Turbo: 38.0 (#200)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| LMArena Longer Query | 1435 | 1254 |

## Writing & Preference

- DeepSeek V4 Flash: 64.0 (#59)
- GPT-4 Turbo: 47.9 (#201)

| Benchmark | DeepSeek V4 Flash | GPT-4 Turbo |
|---|---|---|
| LMArena Text | 1432 | 1272 |
| LMArena Creative Writing | 1403 | 1269 |
| LMArena Multi-Turn | 1449 | 1267 |
| EQ-Bench Creative Writing | 1559 | — |

## FAQ

### Is DeepSeek V4 Flash better than GPT-4 Turbo?

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

### Which is cheaper, DeepSeek V4 Flash or GPT-4 Turbo?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-4 Turbo lists at $10 and $30.

### Is DeepSeek V4 Flash or GPT-4 Turbo better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.6 versus 33.4 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 128K.

### How many benchmarks do DeepSeek V4 Flash and GPT-4 Turbo share?

26 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and GPT-4 Turbo has 36.