# DeepSeek V4 Flash vs GPT-5.2

> DeepSeek V4 Flash and GPT-5.2 score almost the same on the Noometry Index (53.6 vs 54.1), so choose on price, context window or the category you care about most.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-gpt-5-2
- Last updated: 2026-10-11
- Shared benchmarks: 38

## Summary

- They share 38 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 3 categories and GPT-5.2 in 5 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.2 leads 59.3 to 55.4.
- The biggest single-benchmark swing is ProofBench: 56% for DeepSeek V4 Flash and 15% for GPT-5.2.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 400K.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 53.6 | 54.1 |
| Rank | 35 | 34 |
| Context | 1M | 400K |
| Input $/M | $0.15 | $1.75 |
| Output $/M | $0.60 | $14 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Flash: 47.9 (#59)
- GPT-5.2: 51.6 (#37)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena WebDev | 1582 | 1416 |
| WeirdML | 63% | 72.2% |
| LMArena Coding | 1457 | 1447 |
| ALE-Bench | 1,306 | 1,294 |
| SWE-bench Verified | — | 73.8% |
| FrontierCode | 18.8% | — |
| SWE-bench Verified (bash only) | — | 72.8% |
| SWE-bench Multilingual | — | 66.7% |
| SciCode | 49.9% | — |
| GSO | — | 27.4% |
| AlgoTune | — | 2.05 |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- GPT-5.2: 40.2 (#24)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| Terminal-Bench | — | 64.9% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| Remote Labor Index | — | 2.5% |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| DeepResearch Bench | — | 41.1% |
| LMArena Search | — | 1207 |
| METR Time Horizons | — | 75.3% |
| Vending-Bench 2 | — | 3,591 |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#30)
- GPT-5.2: 50.2 (#35)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 61.4% | 52.9% |
| SimpleBench | 61.1% | 45.8% |
| Kagi LLM Benchmark | 52.2% | 73.3% |
| NYT Connections (extended) | 89.6% | 83.6% |
| ARC-AGI-1 | 89% | 86.2% |
| Chess Puzzles | 33% | 49% |
| LMArena Hard Prompts | 1444 | 1445 |
| Mystery Game Puzzles | 34% | 23% |
| DTBench | 90.9% | 90.9% |
| LMCA | 41.7% | 43.9% |
| Epoch Capabilities Index | 154.49 | 153.45 |
| CritPt | 16.6% | — |
| EnigmaEval | — | 10.4% |
| EBR-Bench | — | 23% |
| ForecastBench | — | 60.1 |

## Math

- DeepSeek V4 Flash: 60.3 (#37)
- GPT-5.2: 60.0 (#38)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 67.4% |
| FrontierMath Tier 4 | 24.4% | 31.7% |
| MathArena Final-Answer Competitions | 76.5% | 72% |
| OTIS Mock AIME 2024-2025 | 94.4% | 96.1% |
| ProofBench | 56% | 15% |
| LMArena Math | 1427 | 1440 |
| FrontierMath (Feb 2025 set) | — | 40.7% |
| FrontierMath Tier 4 (v1) | — | 18.8% |

## Knowledge

- DeepSeek V4 Flash: 55.4 (#48)
- GPT-5.2: 59.3 (#32)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 91% | 91.4% |
| SimpleQA Verified | 33.6% | 37.1% |
| LMArena Expert | 1441 | 1445 |
| Humanity's Last Exam | — | 27.8% |
| Vectara Hallucination Rate | — | 8.4% |

## Multimodal

- DeepSeek V4 Flash: —
- GPT-5.2: 51.3 (#7)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena Vision | — | 1268 |
| VPCT | — | 84% |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |

## Multilingual

- DeepSeek V4 Flash: 53.0 (#72)
- GPT-5.2: 53.4 (#67)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1420 | 1425 |
| LMArena Chinese | 1468 | 1460 |
| LMArena French | 1439 | 1455 |
| LMArena German | 1418 | 1448 |
| LMArena Japanese | 1406 | 1420 |
| LMArena Korean | 1384 | 1392 |
| LMArena Russian | 1428 | 1440 |
| LMArena Spanish | 1436 | 1433 |

## Instruction Following

- DeepSeek V4 Flash: 74.9 (#81)
- GPT-5.2: 74.7 (#89)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1421 | 1417 |

## Long Context

- DeepSeek V4 Flash: 43.8 (#85)
- GPT-5.2: 44.0 (#78)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1434 | 1428 |
| CL-bench | — | 18.2% |

## Writing & Preference

- DeepSeek V4 Flash: 63.8 (#61)
- GPT-5.2: 66.8 (#32)

| Benchmark | DeepSeek V4 Flash | GPT-5.2 |
|---|---|---|
| LMArena Text | 1432 | 1439 |
| LMArena Creative Writing | 1403 | 1401 |
| EQ-Bench Creative Writing | 1559 | 1703 |
| LMArena Multi-Turn | 1449 | 1458 |

## FAQ

### Is DeepSeek V4 Flash better than GPT-5.2?

DeepSeek V4 Flash and GPT-5.2 score almost the same on the Noometry Index (53.6 vs 54.1), so choose on price, context window or the category you care about most.

### Which is cheaper, DeepSeek V4 Flash or GPT-5.2?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-5.2 lists at $1.75 and $14.

### Is DeepSeek V4 Flash or GPT-5.2 better for coding?

GPT-5.2 scores higher on coding benchmarks: 51.6 versus 47.9 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 400K.

### How many benchmarks do DeepSeek V4 Flash and GPT-5.2 share?

38 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and GPT-5.2 has 67.