# DeepSeek-V3 vs Gemini 2.5 Flash

> DeepSeek-V3 and Gemini 2.5 Flash score almost the same on the Noometry Index (39.5 vs 39.3), so choose on price, context window or the category you care about most.

- Canonical page: https://noometry.com/compare/deepseek-v3-vs-gemini-2-5-flash
- Last updated: 2026-10-10
- Shared benchmarks: 38

## Summary

- They share 38 benchmarks with published results for both. DeepSeek-V3 scores higher in 4 categories and Gemini 2.5 Flash in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where Gemini 2.5 Flash leads 47.5 to 34.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 73.1% for Gemini 2.5 Flash.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $0.30 / $2.50 for Gemini 2.5 Flash.
- Gemini 2.5 Flash accepts more context: 1.05M tokens versus 164K.
- DeepSeek-V3 has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| Provider | DeepSeek | Google |
| Noometry Index | 39.5 | 39.3 |
| Rank | 166 | 170 |
| Context | 164K | 1.05M |
| Input $/M | $0.24 | $0.30 |
| Output $/M | $0.90 | $2.50 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek-V3: 42.3 (#106)
- Gemini 2.5 Flash: 35.8 (#220)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| Aider Polyglot | 55.1% | 55.1% |
| WeirdML | 36.1% | 41.9% |
| LMArena Coding | 1368 | 1424 |
| SWE-bench Verified (bash only) | — | 28.7% |
| SciCode | 35.8% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| ALE-Bench | — | 661.88 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |

## Agentic & Tool Use

- DeepSeek-V3: —
- Gemini 2.5 Flash: 30.8 (#74)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| Terminal-Bench | — | 17.1% |
| Berkeley Function Calling Leaderboard | — | 56.2% |
| TheAgentCompany | — | 41.1% |
| BALROG | — | 33.5% |
| METR Time Horizons | 49.6% | — |
| Vending-Bench 2 | — | 548.84 |

## Reasoning

- DeepSeek-V3: 20.5 (#236)
- Gemini 2.5 Flash: 18.1 (#286)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| SimpleBench | 27.2% | 41.2% |
| Kagi LLM Benchmark | 52.3% | 56.8% |
| CritPt | 0% | 1.1% |
| LMArena Hard Prompts | 1365 | 1422 |
| DTBench | 64.8% | 76.5% |
| LMCA | 15.5% | 27.5% |
| Epoch Capabilities Index | 135.94 | 143.03 |
| ForecastBench | 59.1 | 60.6 |
| ARC-AGI-2 | — | 2.5% |
| ARC-AGI-1 | — | 33.3% |
| EnigmaEval | — | 2.7% |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |

## Math

- DeepSeek-V3: 32.1 (#219)
- Gemini 2.5 Flash: 39.9 (#98)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 73.1% |
| Omni-MATH | 40.3% | 38.5% |
| LMArena Math | 1373 | 1415 |
| FrontierMath (Feb 2025 set) | 1.7% | 4.8% |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath Tier 4 (v1) | — | 4.2% |

## Knowledge

- DeepSeek-V3: 37.5 (#155)
- Gemini 2.5 Flash: 36.4 (#168)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| MMLU-Pro | 72.3% | 63.9% |
| Confabulations | 26.1% | 16.8% |
| Vectara Hallucination Rate | 6.1% | 7.8% |
| GPQA (HELM) | 53.8% | 39% |
| LMArena Expert | 1351 | 1426 |
| GPQA Diamond | 67.6% | — |
| Humanity's Last Exam | — | 12.1% |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |

## Multimodal

- DeepSeek-V3: —
- Gemini 2.5 Flash: 41.8 (#32)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Vision | — | 1253 |
| GeoBench | — | 76% |
| VPCT | — | 46.2% |
| SpatialViz-Bench | — | 36.9% |

## Multilingual

- DeepSeek-V3: 48.5 (#143)
- Gemini 2.5 Flash: 52.3 (#88)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Non-English | 1358 | 1409 |
| LMArena Chinese | 1391 | 1450 |
| LMArena French | 1385 | 1433 |
| LMArena German | 1374 | 1418 |
| LMArena Japanese | 1333 | 1405 |
| LMArena Korean | 1319 | 1385 |
| LMArena Russian | 1373 | 1415 |
| LMArena Spanish | 1358 | 1421 |

## Instruction Following

- DeepSeek-V3: 72.8 (#130)
- Gemini 2.5 Flash: 75.7 (#54)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| IFEval | 83.2% | 89.8% |
| LMArena Instruction Following | 1345 | 1405 |
| LiveBench Instruction Following | 81.5% | — |

## Long Context

- DeepSeek-V3: 34.0 (#253)
- Gemini 2.5 Flash: 47.5 (#17)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| Fiction.LiveBench | 50% | 77.8% |
| LMArena Longer Query | 1352 | 1419 |

## Writing & Preference

- DeepSeek-V3: 57.4 (#130)
- Gemini 2.5 Flash: 53.8 (#157)

| Benchmark | DeepSeek-V3 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Text | 1375 | 1417 |
| LMArena Creative Writing | 1364 | 1400 |
| Short-Story Creative Writing | 77% | 76.5% |
| EQ-Bench Creative Writing | 1472 | 1137 |
| WildBench | 83% | 81.7% |
| LMArena Multi-Turn | 1389 | 1408 |
| LiveBench Language | 49.1% | — |

## FAQ

### Is DeepSeek-V3 better than Gemini 2.5 Flash?

DeepSeek-V3 and Gemini 2.5 Flash score almost the same on the Noometry Index (39.5 vs 39.3), so choose on price, context window or the category you care about most.

### Which is cheaper, DeepSeek-V3 or Gemini 2.5 Flash?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Gemini 2.5 Flash lists at $0.30 and $2.50.

### Is DeepSeek-V3 or Gemini 2.5 Flash better for coding?

DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 35.8 in the Noometry coding category.

### Which has the bigger context window?

Gemini 2.5 Flash does, with 1.05M tokens against 164K.

### How many benchmarks do DeepSeek-V3 and Gemini 2.5 Flash share?

38 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemini 2.5 Flash has 54.