# Gemma 3 4B vs Qwen2-72B

> Qwen2-72B is the stronger model overall, scoring 29.5 to 27.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-3-4b-vs-qwen2-72b
- Last updated: 2026-10-10
- Shared benchmarks: 14

## Summary

- They share 14 benchmarks with published results for both. Gemma 3 4B scores higher in 6 categories and Qwen2-72B in 3 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen2-72B leads 27.9 to 15.8.
- The biggest single-benchmark swing is GPQA Diamond: 23.2% for Gemma 3 4B and 40.8% for Qwen2-72B.

## Snapshot

| | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| Provider | Google | Alibaba (Qwen) |
| Noometry Index | 27.9 | 29.5 |
| Rank | 320 | 294 |
| Context | 131K | — |
| Input $/M | $0.04 | — |
| Output $/M | $0.08 | — |
| Weights | Open | Open |

## Coding

- Gemma 3 4B: 35.3 (#214)
- Qwen2-72B: 28.7 (#303)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Coding | 1230 | 1196 |
| WeirdML | — | 11.3% |
| BigCodeBench Instruct | — | 38.5% |
| BigCodeBench Complete | — | 54% |

## Agentic & Tool Use

- Gemma 3 4B: 20.7 (#134)
- Qwen2-72B: 16.9 (#138)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | — |
| TheAgentCompany | — | 1.1% |
| METR Time Horizons | — | 29.9% |

## Reasoning

- Gemma 3 4B: 12.7 (#330)
- Qwen2-72B: 22.2 (#184)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Hard Prompts | 1253 | 1191 |
| Epoch Capabilities Index | 116.02 | 125.27 |
| Kagi LLM Benchmark | 25.2% | — |
| Chess Puzzles | 0% | — |
| DTBench | 50.9% | — |
| LMCA | 2.8% | — |

## Math

- Gemma 3 4B: 15.8 (#286)
- Qwen2-72B: 27.9 (#232)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Math | 1239 | 1235 |
| OTIS Mock AIME 2024-2025 | 7.5% | — |
| MATH Level 5 | — | 39.1% |

## Knowledge

- Gemma 3 4B: 11.8 (#294)
- Qwen2-72B: 21.2 (#270)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| GPQA Diamond | 23.2% | 40.8% |
| LMArena Expert | 1223 | 1171 |
| Vectara Hallucination Rate | 6.4% | — |
| MMLU | — | 82.4% |

## Multilingual

- Gemma 3 4B: 42.7 (#189)
- Qwen2-72B: 36.0 (#239)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Non-English | 1273 | 1176 |
| LMArena German | 1281 | 1151 |
| LMArena Russian | 1294 | 1169 |
| LMArena Chinese | — | 1240 |
| LMArena French | — | 1169 |
| LMArena Japanese | — | 1111 |
| LMArena Korean | — | 1083 |
| LMArena Spanish | — | 1168 |

## Instruction Following

- Gemma 3 4B: 65.4 (#219)
- Qwen2-72B: 62.0 (#236)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Instruction Following | 1239 | 1181 |

## Long Context

- Gemma 3 4B: 38.6 (#189)
- Qwen2-72B: 36.1 (#230)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Longer Query | 1274 | 1193 |

## Writing & Preference

- Gemma 3 4B: 42.2 (#234)
- Qwen2-72B: 41.0 (#236)

| Benchmark | Gemma 3 4B | Qwen2-72B |
|---|---|---|
| LMArena Text | 1291 | 1203 |
| LMArena Creative Writing | 1270 | 1181 |
| LMArena Multi-Turn | 1255 | 1195 |
| EQ-Bench Creative Writing | 1068 | — |

## FAQ

### Is Gemma 3 4B better than Qwen2-72B?

Qwen2-72B is the stronger model overall, scoring 29.5 to 27.9 on the Noometry Index.

### Is Gemma 3 4B or Qwen2-72B better for coding?

Gemma 3 4B scores higher on coding benchmarks: 35.3 versus 28.7 in the Noometry coding category.

### How many benchmarks do Gemma 3 4B and Qwen2-72B share?

14 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Qwen2-72B has 26.