# Gemma 7B vs Qwen2.5 72B Instruct

> Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 30.0 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-7b-vs-qwen2-5-72b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 21

## Summary

- They share 21 benchmarks with published results for both. Gemma 7B scores higher in 2 categories and Qwen2.5 72B Instruct in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen2.5 72B Instruct leads 46.7 to 27.1.

## Snapshot

| | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| Provider | Google | Alibaba (Qwen) |
| Noometry Index | 30.0 | 31.9 |
| Rank | 299 | 267 |
| Context | — | 131K |
| Input $/M | — | $1.40 |
| Output $/M | — | $5.60 |
| Weights | Open | Open |

## Coding

- Gemma 7B: 30.5 (#294)
- Qwen2.5 72B Instruct: 33.2 (#260)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Coding | 1048 | 1292 |
| WeirdML | — | 16% |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| HumanEval+ | 28.7% | — |
| MBPP+ | 43.4% | — |

## Agentic & Tool Use

- Gemma 7B: —
- Qwen2.5 72B Instruct: 22.1 (#133)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |

## Reasoning

- Gemma 7B: 19.9 (#249)
- Qwen2.5 72B Instruct: 22.3 (#199)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1042 | 1271 |
| BIG-Bench Hard | 55.1% | 79.8% |
| Epoch Capabilities Index | 111.99 | 129 |
| HellaSwag | 82.2% | 84.8% |
| PIQA | 81.2% | 82.6% |
| WinoGrande | 79% | 82.3% |
| DTBench | — | 62.9% |
| LMCA | — | 13.4% |
| Adversarial NLI | 48.7% | — |
| ForecastBench | — | 57.5 |

## Math

- Gemma 7B: 31.2 (#228)
- Qwen2.5 72B Instruct: 19.3 (#287)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Math | 1066 | 1283 |
| OTIS Mock AIME 2024-2025 | — | 8.1% |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
| GSM8K | 46.4% | — |

## Knowledge

- Gemma 7B: 27.3 (#252)
- Qwen2.5 72B Instruct: 27.0 (#253)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Expert | 1001 | 1245 |
| ARC (AI2) Challenge | 78.3% | 94.5% |
| MMLU | 66.1% | 85.3% |
| TriviaQA | 72.3% | 71.9% |
| GPQA Diamond | — | 49.1% |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| BoolQ | 83.2% | — |
| OpenBookQA | 78.6% | — |

## Multilingual

- Gemma 7B: 25.1 (#287)
- Qwen2.5 72B Instruct: 41.0 (#213)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 999 | 1252 |
| LMArena Chinese | 1035 | 1272 |
| LMArena French | 1025 | 1280 |
| LMArena Russian | 993 | 1264 |
| LMArena German | — | 1234 |
| LMArena Japanese | — | 1180 |
| LMArena Korean | — | 1188 |
| LMArena Spanish | — | 1256 |

## Instruction Following

- Gemma 7B: 51.5 (#295)
- Qwen2.5 72B Instruct: 65.5 (#221)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1017 | 1254 |
| IFEval | — | 80.6% |

## Long Context

- Gemma 7B: 31.1 (#282)
- Qwen2.5 72B Instruct: 38.9 (#188)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1022 | 1282 |

## Writing & Preference

- Gemma 7B: 27.1 (#302)
- Qwen2.5 72B Instruct: 46.7 (#215)

| Benchmark | Gemma 7B | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1056 | 1269 |
| LMArena Creative Writing | 1024 | 1221 |
| LMArena Multi-Turn | 963 | 1272 |
| WildBench | — | 80.2% |

## FAQ

### Is Gemma 7B better than Qwen2.5 72B Instruct?

Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 30.0 on the Noometry Index.

### Is Gemma 7B or Qwen2.5 72B Instruct better for coding?

Qwen2.5 72B Instruct scores higher on coding benchmarks: 33.2 versus 30.5 in the Noometry coding category.

### How many benchmarks do Gemma 7B and Qwen2.5 72B Instruct share?

21 benchmarks have published results for both models. Gemma 7B has 27 scored results on Noometry and Qwen2.5 72B Instruct has 43.