# DeepSeek-V2.5 (Sep 2024) vs Gemma 4 31B IT

> Gemma 4 31B IT is the stronger model overall, scoring 43.2 to 37.0 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v2-5-vs-gemma-4-31b-it
- Last updated: 2026-10-10
- Shared benchmarks: 14

## Summary

- They share 14 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 0 categories and Gemma 4 31B IT in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in multilingual, where Gemma 4 31B IT leads 54.0 to 42.7.

## Snapshot

| | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| Provider | DeepSeek | Google |
| Noometry Index | 37.0 | 43.2 |
| Rank | 197 | 85 |
| Context | — | 262K |
| Input $/M | — | $0.09 |
| Output $/M | — | $0.34 |
| Weights | Open | Open |

## Coding

- DeepSeek-V2.5 (Sep 2024): 31.3 (#274)
- Gemma 4 31B IT: 41.9 (#98)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Coding | 1310 | 1459 |
| Aider Polyglot | 17.8% | — |
| LMArena WebDev | — | 1364 |
| SciCode | — | 43.4% |
| WeirdML | — | 52.3% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 925.5 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |

## Reasoning

- DeepSeek-V2.5 (Sep 2024): 24.5 (#145)
- Gemma 4 31B IT: 26.4 (#119)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1448 |
| Kagi LLM Benchmark | — | 63.5% |
| NYT Connections (extended) | — | 70.6% |
| CritPt | — | 1.4% |
| Chess Puzzles | — | 5% |
| Thematic Generalization | — | 53% |
| DTBench | — | 82.7% |
| LMCA | — | 39.3% |
| Surface Evolver Bench | — | 30.6% |
| Epoch Capabilities Index | — | 142.74 |

## Math

- DeepSeek-V2.5 (Sep 2024): 33.3 (#177)
- Gemma 4 31B IT: 41.3 (#72)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Math | 1289 | 1464 |
| OTIS Mock AIME 2024-2025 | — | 73.3% |

## Knowledge

- DeepSeek-V2.5 (Sep 2024): 34.2 (#190)
- Gemma 4 31B IT: 37.9 (#141)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Expert | 1266 | 1465 |
| GPQA Diamond | — | 75.8% |
| SimpleQA Verified | — | 10.4% |
| Vectara Hallucination Rate | — | 7.4% |

## Multimodal

- DeepSeek-V2.5 (Sep 2024): —
- Gemma 4 31B IT: 41.0 (#32)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Vision | — | 1278 |
| LMArena Document | — | 1425 |

## Multilingual

- DeepSeek-V2.5 (Sep 2024): 42.7 (#188)
- Gemma 4 31B IT: 54.0 (#52)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Non-English | 1273 | 1431 |
| LMArena Chinese | 1317 | 1478 |
| LMArena French | 1288 | 1436 |
| LMArena Russian | 1289 | 1462 |
| LMArena Spanish | 1248 | 1443 |
| LMArena German | 1258 | — |
| LMArena Japanese | 1228 | — |
| LMArena Korean | 1209 | — |

## Instruction Following

- DeepSeek-V2.5 (Sep 2024): 67.7 (#189)
- Gemma 4 31B IT: 75.6 (#58)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Instruction Following | 1280 | 1433 |

## Long Context

- DeepSeek-V2.5 (Sep 2024): 39.5 (#169)
- Gemma 4 31B IT: 44.2 (#67)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Longer Query | 1301 | 1446 |

## Writing & Preference

- DeepSeek-V2.5 (Sep 2024): 50.0 (#182)
- Gemma 4 31B IT: 60.7 (#92)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Gemma 4 31B IT |
|---|---|---|
| LMArena Text | 1294 | 1443 |
| LMArena Creative Writing | 1285 | 1416 |
| LMArena Multi-Turn | 1297 | 1452 |
| EQ-Bench Creative Writing | — | 1368 |
| EQ-Bench 4 | — | 1120 |

## FAQ

### Is DeepSeek-V2.5 (Sep 2024) better than Gemma 4 31B IT?

Gemma 4 31B IT is the stronger model overall, scoring 43.2 to 37.0 on the Noometry Index.

### Is DeepSeek-V2.5 (Sep 2024) or Gemma 4 31B IT better for coding?

Gemma 4 31B IT scores higher on coding benchmarks: 41.9 versus 31.3 in the Noometry coding category.

### How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Gemma 4 31B IT share?

14 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Gemma 4 31B IT has 35.