# DeepSeek V4 Flash vs o4-mini

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 41.6 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-flash-vs-o4-mini
- Last updated: 2026-10-11
- Shared benchmarks: 34

## Summary

- They share 34 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 6 categories and o4-mini in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 24.6.
- The biggest single-benchmark swing is ARC-AGI-2: 61.4% for DeepSeek V4 Flash and 6.1% for o4-mini.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $1.10 / $4.40 for o4-mini.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 200K.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 53.6 | 41.6 |
| Rank | 35 | 132 |
| Context | 1M | 200K |
| Input $/M | $0.15 | $1.10 |
| Output $/M | $0.60 | $4.40 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek V4 Flash: 47.9 (#59)
- o4-mini: 40.9 (#127)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| WeirdML | 63% | 52.6% |
| LMArena Coding | 1457 | 1368 |
| ALE-Bench | 1,306 | 826.17 |
| FrontierCode | 18.8% | — |
| SWE-bench Verified (bash only) | — | 45% |
| Aider Polyglot | — | 72% |
| LMArena WebDev | 1582 | — |
| SciCode | 49.9% | — |
| GSO | — | 3.6% |
| CadEval | — | 62% |
| AlgoTune | — | 1.72 |

## Agentic & Tool Use

- DeepSeek V4 Flash: —
- o4-mini: 32.6 (#61)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 53.2% |
| GDPval | — | 25.3% |
| METR Time Horizons | — | 63.9% |

## Reasoning

- DeepSeek V4 Flash: 53.7 (#30)
- o4-mini: 24.6 (#162)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| ARC-AGI-2 | 61.4% | 6.1% |
| SimpleBench | 61.1% | 38.7% |
| Kagi LLM Benchmark | 52.2% | 67.6% |
| ARC-AGI-1 | 89% | 58.7% |
| CritPt | 16.6% | 0.6% |
| Chess Puzzles | 33% | 26% |
| LMArena Hard Prompts | 1444 | 1351 |
| Mystery Game Puzzles | 34% | 5% |
| DTBench | 90.9% | 77.6% |
| LMCA | 41.7% | 26.5% |
| Epoch Capabilities Index | 154.49 | 145.64 |
| NYT Connections (extended) | 89.6% | — |
| EnigmaEval | — | 9.2% |
| ForecastBench | — | 61.8 |

## Math

- DeepSeek V4 Flash: 60.3 (#37)
- o4-mini: 40.8 (#89)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.5% | 36.1% |
| FrontierMath Tier 4 | 24.4% | 4.9% |
| OTIS Mock AIME 2024-2025 | 94.4% | 81.7% |
| LMArena Math | 1427 | 1389 |
| MathArena Final-Answer Competitions | 76.5% | — |
| ProofBench | 56% | — |
| Omni-MATH | — | 72% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 24.8% |
| FrontierMath Tier 4 (v1) | — | 6.3% |

## Knowledge

- DeepSeek V4 Flash: 55.4 (#48)
- o4-mini: 43.6 (#91)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| GPQA Diamond | 91% | 79.6% |
| SimpleQA Verified | 33.6% | 19.6% |
| LMArena Expert | 1441 | 1343 |
| Humanity's Last Exam | — | 18.1% |
| MMLU-Pro | — | 82% |
| Confabulations | — | 15.8% |
| Vectara Hallucination Rate | — | 18.6% |
| GPQA (HELM) | — | 73.5% |

## Multimodal

- DeepSeek V4 Flash: —
- o4-mini: 40.2 (#49)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| LMArena Vision | — | 1194 |
| GeoBench | — | 64% |
| VPCT | — | 57.5% |

## Multilingual

- DeepSeek V4 Flash: 53.0 (#72)
- o4-mini: 47.0 (#154)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| LMArena Non-English | 1420 | 1337 |
| LMArena Chinese | 1468 | 1354 |
| LMArena French | 1439 | 1364 |
| LMArena German | 1418 | 1336 |
| LMArena Japanese | 1406 | 1308 |
| LMArena Korean | 1384 | 1312 |
| LMArena Russian | 1428 | 1334 |
| LMArena Spanish | 1436 | 1347 |

## Instruction Following

- DeepSeek V4 Flash: 74.9 (#81)
- o4-mini: 75.2 (#68)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| LMArena Instruction Following | 1421 | 1321 |
| IFEval | — | 92.8% |

## Long Context

- DeepSeek V4 Flash: 43.8 (#85)
- o4-mini: 45.5 (#33)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| LMArena Longer Query | 1434 | 1315 |
| Fiction.LiveBench | — | 77.8% |

## Writing & Preference

- DeepSeek V4 Flash: 63.8 (#61)
- o4-mini: 54.0 (#152)

| Benchmark | DeepSeek V4 Flash | o4-mini |
|---|---|---|
| LMArena Text | 1432 | 1353 |
| LMArena Creative Writing | 1403 | 1294 |
| LMArena Multi-Turn | 1449 | 1350 |
| Short-Story Creative Writing | — | 75% |
| EQ-Bench Creative Writing | 1559 | — |
| WildBench | — | 85.4% |

## FAQ

### Is DeepSeek V4 Flash better than o4-mini?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 41.6 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Flash or o4-mini?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; o4-mini lists at $1.10 and $4.40.

### Is DeepSeek V4 Flash or o4-mini better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 40.9 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Flash does, with 1M tokens against 200K.

### How many benchmarks do DeepSeek V4 Flash and o4-mini share?

34 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and o4-mini has 60.