# DeepSeek V4 Pro vs Gemma 2 27B

> DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 29.2 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v4-pro-vs-gemma-2-27b
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 8 categories and Gemma 2 27B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.6 to 10.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 98.6% for DeepSeek V4 Pro and 1.4% for Gemma 2 27B.
- DeepSeek V4 Pro is cheaper at $0.43 / $0.87 per million input/output tokens, against $0.65 / $0.65 for Gemma 2 27B.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 8K.

## Snapshot

| | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| Provider | DeepSeek | Google |
| Noometry Index | 54.2 | 29.2 |
| Rank | 28 | 302 |
| Context | 1M | 8K |
| Input $/M | $0.43 | $0.65 |
| Output $/M | $0.87 | $0.65 |
| Weights | Open | Open |

## Coding

- DeepSeek V4 Pro: 52.2 (#31)
- Gemma 2 27B: 33.5 (#243)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Coding | 1470 | 1211 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| WeirdML | 66.2% | — |
| BigCodeBench Instruct | — | 42.8% |
| LiveBench Coding | — | 36% |
| BigCodeBench Complete | — | 52.5% |
| ALE-Bench | 1,403 | — |

## Agentic & Tool Use

- DeepSeek V4 Pro: 32.7 (#54)
- Gemma 2 27B: —

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| APEX-Agents | 47.3% | — |
| Vending-Bench 2 | 3,285 | — |

## Reasoning

- DeepSeek V4 Pro: 56.4 (#22)
- Gemma 2 27B: 14.8 (#313)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Hard Prompts | 1461 | 1199 |
| DTBench | 93.9% | 48% |
| LMCA | 45.5% | 7.1% |
| Epoch Capabilities Index | 155.39 | 122.06 |
| ARC-AGI-2 | 61.3% | — |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| Chess Puzzles | 47% | — |
| LiveBench Reasoning | — | 28.1% |
| Mystery Game Puzzles | 43% | — |
| LiveBench Data Analysis | — | 47.9% |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
| LiveBench | — | 38.2% |

## Math

- DeepSeek V4 Pro: 64.6 (#24)
- Gemma 2 27B: 10.0 (#306)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.6% | 1.4% |
| LMArena Math | 1455 | 1212 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| LiveBench Math | — | 26.5% |
| MATH Level 5 | — | 27.9% |

## Knowledge

- DeepSeek V4 Pro: 59.5 (#28)
- Gemma 2 27B: 18.9 (#275)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| GPQA Diamond | 91.7% | 36.5% |
| LMArena Expert | 1463 | 1172 |
| SimpleQA Verified | 52.9% | — |
| Confabulations | — | 27.1% |
| Vectara Hallucination Rate | 8.6% | — |
| MMLU | — | 75.7% |

## Multilingual

- DeepSeek V4 Pro: 54.5 (#44)
- Gemma 2 27B: 38.8 (#221)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Non-English | 1438 | 1217 |
| LMArena Chinese | 1486 | 1221 |
| LMArena French | 1473 | 1246 |
| LMArena German | 1458 | 1209 |
| LMArena Japanese | 1448 | 1175 |
| LMArena Korean | 1452 | 1174 |
| LMArena Russian | 1452 | 1234 |
| LMArena Spanish | 1457 | 1228 |

## Instruction Following

- DeepSeek V4 Pro: 76.3 (#44)
- Gemma 2 27B: 60.7 (#244)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Instruction Following | 1448 | 1206 |
| LiveBench Instruction Following | — | 58.1% |

## Long Context

- DeepSeek V4 Pro: 45.0 (#49)
- Gemma 2 27B: 37.3 (#213)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Longer Query | 1459 | 1231 |
| CL-bench Life | 13.5% | — |

## Writing & Preference

- DeepSeek V4 Pro: 65.6 (#45)
- Gemma 2 27B: 44.4 (#220)

| Benchmark | DeepSeek V4 Pro | Gemma 2 27B |
|---|---|---|
| LMArena Text | 1451 | 1231 |
| LMArena Creative Writing | 1446 | 1241 |
| LMArena Multi-Turn | 1467 | 1224 |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |
| LiveBench Language | — | 32.6% |

## FAQ

### Is DeepSeek V4 Pro better than Gemma 2 27B?

DeepSeek V4 Pro is the stronger model overall, scoring 54.2 to 29.2 on the Noometry Index.

### Which is cheaper, DeepSeek V4 Pro or Gemma 2 27B?

DeepSeek V4 Pro is cheaper. It lists at $0.43 per million input tokens and $0.87 per million output tokens; Gemma 2 27B lists at $0.65 and $0.65.

### Is DeepSeek V4 Pro or Gemma 2 27B better for coding?

DeepSeek V4 Pro scores higher on coding benchmarks: 52.2 versus 33.5 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek V4 Pro does, with 1M tokens against 8K.

### How many benchmarks do DeepSeek V4 Pro and Gemma 2 27B share?

22 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Gemma 2 27B has 34.