# Claude Sonnet 4 vs DeepSeek V4 Flash

> DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 40.8 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-vs-deepseek-v4-flash
- Last updated: 2026-10-10
- Shared benchmarks: 31

## Summary

- They share 31 benchmarks with published results for both. Claude Sonnet 4 scores higher in 0 categories and DeepSeek V4 Flash in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 22.9.
- The biggest single-benchmark swing is ARC-AGI-2: 5.9% for Claude Sonnet 4 and 61.4% for DeepSeek V4 Flash.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 200K.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.

## Snapshot

| | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 40.8 | 53.6 |
| Rank | 145 | 35 |
| Context | 200K | 1M |
| Input $/M | $3 | $0.15 |
| Output $/M | $15 | $0.60 |
| Weights | Proprietary | Open |

## Coding

- Claude Sonnet 4: 43.5 (#88)
- DeepSeek V4 Flash: 47.9 (#59)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| SciCode | 40% | 49.9% |
| WeirdML | 46.1% | 63% |
| LMArena Coding | 1414 | 1457 |
| ALE-Bench | 655.35 | 1,306 |
| FrontierCode | — | 18.8% |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| LMArena WebDev | — | 1582 |
| GSO | 4.9% | — |

## Agentic & Tool Use

- Claude Sonnet 4: 38.5 (#31)
- DeepSeek V4 Flash: —

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |

## Reasoning

- Claude Sonnet 4: 22.9 (#187)
- DeepSeek V4 Flash: 53.7 (#30)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| ARC-AGI-2 | 5.9% | 61.4% |
| SimpleBench | 45.5% | 61.1% |
| Kagi LLM Benchmark | 73% | 52.2% |
| ARC-AGI-1 | 40% | 89% |
| CritPt | 0.3% | 16.6% |
| LMArena Hard Prompts | 1372 | 1444 |
| DTBench | 77.1% | 90.9% |
| LMCA | 29% | 41.7% |
| Epoch Capabilities Index | 141.69 | 154.49 |
| NYT Connections (extended) | — | 89.6% |
| Chess Puzzles | — | 33% |
| EnigmaEval | 3.1% | — |
| Mystery Game Puzzles | — | 34% |
| ForecastBench | 60.2 | — |

## Math

- Claude Sonnet 4: 43.3 (#80)
- DeepSeek V4 Flash: 60.3 (#37)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 94.4% |
| LMArena Math | 1375 | 1427 |
| FrontierMath (Tiers 1-3) | — | 57.5% |
| FrontierMath Tier 4 | — | 24.4% |
| MathArena Final-Answer Competitions | — | 76.5% |
| ProofBench | — | 56% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |

## Knowledge

- Claude Sonnet 4: 41.8 (#108)
- DeepSeek V4 Flash: 55.4 (#48)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| GPQA Diamond | 79.2% | 91% |
| LMArena Expert | 1372 | 1441 |
| Humanity's Last Exam | 7.8% | — |
| SimpleQA Verified | — | 33.6% |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |

## Multimodal

- Claude Sonnet 4: 26.2 (#121)
- DeepSeek V4 Flash: —

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Vision | 1191 | — |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |

## Multilingual

- Claude Sonnet 4: 46.7 (#156)
- DeepSeek V4 Flash: 53.0 (#72)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Non-English | 1333 | 1420 |
| LMArena Chinese | 1350 | 1468 |
| LMArena French | 1363 | 1439 |
| LMArena German | 1331 | 1418 |
| LMArena Japanese | 1302 | 1406 |
| LMArena Korean | 1291 | 1384 |
| LMArena Russian | 1355 | 1428 |
| LMArena Spanish | 1357 | 1436 |

## Instruction Following

- Claude Sonnet 4: 71.7 (#145)
- DeepSeek V4 Flash: 74.9 (#81)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Instruction Following | 1376 | 1421 |
| IFEval | 84% | — |

## Long Context

- Claude Sonnet 4: 33.7 (#259)
- DeepSeek V4 Flash: 43.8 (#85)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Longer Query | 1398 | 1434 |
| Fiction.LiveBench | 46.9% | — |

## Writing & Preference

- Claude Sonnet 4: 57.1 (#132)
- DeepSeek V4 Flash: 63.8 (#61)

| Benchmark | Claude Sonnet 4 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Text | 1351 | 1432 |
| LMArena Creative Writing | 1345 | 1403 |
| EQ-Bench Creative Writing | 1483 | 1559 |
| LMArena Multi-Turn | 1376 | 1449 |
| Short-Story Creative Writing | 81.4% | — |
| WildBench | 83.8% | — |

## FAQ

### Is Claude Sonnet 4 better than DeepSeek V4 Flash?

DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 40.8 on the Noometry Index.

### Which is cheaper, Claude Sonnet 4 or DeepSeek V4 Flash?

DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Claude Sonnet 4 lists at $3 and $15.

### Is Claude Sonnet 4 or DeepSeek V4 Flash better for coding?

DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 43.5 in the Noometry coding category.

### Which has the bigger context window?

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

### How many benchmarks do Claude Sonnet 4 and DeepSeek V4 Flash share?

31 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and DeepSeek V4 Flash has 41.