# DeepSeek-V3.1 vs Gemini 2.5 Flash

> DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 39.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v3-1-vs-gemini-2-5-flash
- Last updated: 2026-10-10
- Shared benchmarks: 27

## Summary

- They share 27 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 4 categories and Gemini 2.5 Flash in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where Gemini 2.5 Flash leads 47.5 to 36.3.
- The biggest single-benchmark swing is Fiction.LiveBench: 52.8% for DeepSeek-V3.1 and 77.8% for Gemini 2.5 Flash.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 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-V3.1 has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| Provider | DeepSeek | Google |
| Noometry Index | 42.8 | 39.3 |
| Rank | 108 | 170 |
| Context | 164K | 1.05M |
| Input $/M | $0.25 | $0.30 |
| Output $/M | $0.95 | $2.50 |
| Weights | Open | Proprietary |

## Coding

- DeepSeek-V3.1: 40.3 (#144)
- Gemini 2.5 Flash: 35.8 (#220)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| WeirdML | 38.4% | 41.9% |
| LMArena Coding | 1417 | 1424 |
| SWE-bench Verified (bash only) | — | 28.7% |
| Aider Polyglot | — | 55.1% |
| ALE-Bench | — | 661.88 |

## Agentic & Tool Use

- DeepSeek-V3.1: —
- Gemini 2.5 Flash: 30.8 (#74)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| Terminal-Bench | — | 17.1% |
| Berkeley Function Calling Leaderboard | — | 56.2% |
| TheAgentCompany | — | 41.1% |
| BALROG | — | 33.5% |
| Vending-Bench 2 | — | 548.84 |

## Reasoning

- DeepSeek-V3.1: 27.9 (#110)
- Gemini 2.5 Flash: 18.1 (#286)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| SimpleBench | 40% | 41.2% |
| Kagi LLM Benchmark | 53.2% | 56.8% |
| LMArena Hard Prompts | 1417 | 1422 |
| DTBench | 82.7% | 76.5% |
| LMCA | 24.3% | 27.5% |
| Epoch Capabilities Index | 139.92 | 143.03 |
| ForecastBench | 58 | 60.6 |
| ARC-AGI-2 | — | 2.5% |
| ARC-AGI-1 | — | 33.3% |
| CritPt | — | 1.1% |
| EnigmaEval | — | 2.7% |

## Math

- DeepSeek-V3.1: 38.9 (#122)
- Gemini 2.5 Flash: 39.9 (#98)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Math | 1420 | 1415 |
| OTIS Mock AIME 2024-2025 | — | 73.1% |
| Omni-MATH | — | 38.5% |
| FrontierMath (Feb 2025 set) | — | 4.8% |
| FrontierMath Tier 4 (v1) | — | 4.2% |

## Knowledge

- DeepSeek-V3.1: 43.7 (#90)
- Gemini 2.5 Flash: 36.4 (#168)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | 7.8% |
| LMArena Expert | 1405 | 1426 |
| Humanity's Last Exam | — | 12.1% |
| MMLU-Pro | — | 63.9% |
| Confabulations | — | 16.8% |
| GPQA (HELM) | — | 39% |

## Multimodal

- DeepSeek-V3.1: —
- Gemini 2.5 Flash: 41.8 (#32)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Vision | — | 1253 |
| GeoBench | — | 76% |
| VPCT | — | 46.2% |
| SpatialViz-Bench | — | 36.9% |

## Multilingual

- DeepSeek-V3.1: 51.6 (#106)
- Gemini 2.5 Flash: 52.3 (#88)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Non-English | 1400 | 1409 |
| LMArena Chinese | 1469 | 1450 |
| LMArena French | 1447 | 1433 |
| LMArena German | 1411 | 1418 |
| LMArena Japanese | 1378 | 1405 |
| LMArena Korean | 1337 | 1385 |
| LMArena Russian | 1405 | 1415 |
| LMArena Spanish | 1431 | 1421 |

## Instruction Following

- DeepSeek-V3.1: 73.9 (#110)
- Gemini 2.5 Flash: 75.7 (#54)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Instruction Following | 1400 | 1405 |
| IFEval | — | 89.8% |

## Long Context

- DeepSeek-V3.1: 36.3 (#232)
- Gemini 2.5 Flash: 47.5 (#17)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| Fiction.LiveBench | 52.8% | 77.8% |
| LMArena Longer Query | 1422 | 1419 |

## Writing & Preference

- DeepSeek-V3.1: 60.3 (#98)
- Gemini 2.5 Flash: 53.8 (#157)

| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| LMArena Text | 1420 | 1417 |
| LMArena Creative Writing | 1401 | 1400 |
| EQ-Bench Creative Writing | 1436 | 1137 |
| LMArena Multi-Turn | 1408 | 1408 |
| Short-Story Creative Writing | — | 76.5% |
| WildBench | — | 81.7% |

## FAQ

### Is DeepSeek-V3.1 better than Gemini 2.5 Flash?

DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 39.3 on the Noometry Index.

### Which is cheaper, DeepSeek-V3.1 or Gemini 2.5 Flash?

DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; Gemini 2.5 Flash lists at $0.30 and $2.50.

### Is DeepSeek-V3.1 or Gemini 2.5 Flash better for coding?

DeepSeek-V3.1 scores higher on coding benchmarks: 40.3 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-V3.1 and Gemini 2.5 Flash share?

27 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Gemini 2.5 Flash has 54.