Model comparison
Gemma 3 4B vs Qwen2.5-Coder-32B
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 28.1 on the Noometry Index. Gemma 3 4B costs 15× less per token, which makes it the better buy when Qwen2.5-Coder-32B's lead doesn't matter for your workload.
Last verified . 12 shared benchmarks.
Google
Alibaba (Qwen)Summary
- They share 12 benchmarks with published results for both. Gemma 3 4B scores higher in 5 categories and Qwen2.5-Coder-32B in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen2.5-Coder-32B leads 33.4 to 11.8.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.66 / $1 for Qwen2.5-Coder-32B.
- Gemma 3 4B accepts more context: 131K tokens versus 33K.
Side by side
| Gemma 3 4B | Qwen2.5-Coder-32B | |
|---|---|---|
| Provider | Alibaba (Qwen) | |
| Noometry Index | 28.1 | 33.4 |
| Released | 2025-03-12 | 2024-09-18 |
| Weights | Open | Open |
| Context window | 131K | 33K |
| Max output | 4K | 29K |
| Input $ / M tokens | $0.04 | $0.66 |
| Output $ / M tokens | $0.08 | $1 |
| Results tracked | 22 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 3 4B leads
Gemma 3 4B: 35.9 (#215), Qwen2.5-Coder-32B: 22.6 (#333)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Coding | 1230 | 1276 |
| SWE-bench Verified (bash only) | — | 9% |
| Aider Polyglot | — | 16.4% |
| BigCodeBench Instruct | — | 49% |
| LiveBench Coding | — | 56.9% |
| BigCodeBench Complete | — | 58% |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 77% |
Agentic & Tool Use Not comparable
Gemma 3 4B: 20.9 (#142), Qwen2.5-Coder-32B: —
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | — |
Reasoning Qwen2.5-Coder-32B leads
Gemma 3 4B: 13.2 (#335), Qwen2.5-Coder-32B: 21.2 (#225)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Hard Prompts | 1253 | 1251 |
| Epoch Capabilities Index | 116.02 | 119.49 |
| Kagi LLM Benchmark | 25.2% | — |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | — | 42.1% |
| DTBench | 50.9% | — |
| LiveBench Data Analysis | — | 49.9% |
| LMCA | 2.8% | — |
| HellaSwag | — | 83% |
| LiveBench | — | 46.2% |
| WinoGrande | — | 80.8% |
Math Qwen2.5-Coder-32B leads
Gemma 3 4B: 16.8 (#292), Qwen2.5-Coder-32B: 33.3 (#204)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Math | 1239 | 1251 |
| OTIS Mock AIME 2024-2025 | 7.5% | — |
| LiveBench Math | — | 46.6% |
| GSM8K | — | 93% |
Knowledge Qwen2.5-Coder-32B leads
Gemma 3 4B: 11.8 (#299), Qwen2.5-Coder-32B: 33.4 (#203)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Expert | 1223 | 1221 |
| GPQA Diamond | 23.2% | — |
| Vectara Hallucination Rate | 6.4% | — |
| ARC (AI2) Challenge | — | 70.5% |
| MMLU | — | 79.1% |
Multilingual Gemma 3 4B leads
Gemma 3 4B: 42.5 (#194), Qwen2.5-Coder-32B: 37.8 (#235)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Non-English | 1273 | 1205 |
| LMArena Russian | 1294 | 1228 |
| LMArena Chinese | — | 1222 |
| LMArena German | 1281 | — |
Instruction Following Gemma 3 4B leads
Gemma 3 4B: 65.2 (#225), Qwen2.5-Coder-32B: 61.4 (#245)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Instruction Following | 1239 | 1223 |
| LiveBench Instruction Following | — | 58.7% |
Long Context Too close to call
Gemma 3 4B: 38.7 (#194), Qwen2.5-Coder-32B: 38.0 (#208)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Longer Query | 1273 | 1251 |
Writing & Preference Too close to call
Gemma 3 4B: 42.0 (#239), Qwen2.5-Coder-32B: 41.6 (#240)
| Benchmark | Gemma 3 4B | Qwen2.5-Coder-32B |
|---|---|---|
| LMArena Text | 1291 | 1230 |
| LMArena Creative Writing | 1271 | 1174 |
| LMArena Multi-Turn | 1255 | 1222 |
| EQ-Bench Creative Writing | 1068 | — |
| LiveBench Language | — | 23.3% |
Frequently asked questions
Is Gemma 3 4B better than Qwen2.5-Coder-32B?
Qwen2.5-Coder-32B is the stronger model overall, scoring 33.4 to 28.1 on the Noometry Index. Gemma 3 4B costs 15× less per token, which makes it the better buy when Qwen2.5-Coder-32B's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or Qwen2.5-Coder-32B?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Qwen2.5-Coder-32B lists at $0.66 and $1.
Is Gemma 3 4B or Qwen2.5-Coder-32B better for coding?
Gemma 3 4B scores higher on coding benchmarks: 35.9 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 4B does, with 131K tokens against 33K.
How many benchmarks do Gemma 3 4B and Qwen2.5-Coder-32B share?
12 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Qwen2.5-Coder-32B has 31.