Model comparison
Gemma 4 31B IT vs Qwen2.5 72B Instruct
Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 31.9 on the Noometry Index.
Last verified . 20 shared benchmarks.
Google
Alibaba (Qwen)Summary
- They share 20 benchmarks with published results for both. Gemma 4 31B IT scores higher in 8 categories and Qwen2.5 72B Instruct in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 4 31B IT leads 43.2 to 19.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 73.3% for Gemma 4 31B IT and 8.1% for Qwen2.5 72B Instruct.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
- Gemma 4 31B IT accepts more context: 262K tokens versus 131K.
Side by side
| Gemma 4 31B IT | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Alibaba (Qwen) | |
| Noometry Index | 43.5 | 31.9 |
| Released | 2026-04-02 | 2024-09 |
| Weights | Open | Open |
| Context window | 262K | 131K |
| Max output | 33K | 8K |
| Input $ / M tokens | $0.09 | $1.40 |
| Output $ / M tokens | $0.34 | $5.60 |
| Results tracked | 35 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemma 4 31B IT leads
Gemma 4 31B IT: 42.3 (#108), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 52.3% | 16% |
| LMArena Coding | 1459 | 1292 |
| LMArena WebDev | 1366 | — |
| SciCode | 43.4% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 925.5 | — |
Agentic & Tool Use Not comparable
Gemma 4 31B IT: —, Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Gemma 4 31B IT leads
Gemma 4 31B IT: 27.2 (#122), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1448 | 1271 |
| DTBench | 82.7% | 62.9% |
| LMCA | 39.3% | 13.4% |
| Epoch Capabilities Index | 142.74 | 129 |
| Kagi LLM Benchmark | 63.5% | — |
| NYT Connections (extended) | 70.6% | — |
| CritPt | 1.4% | — |
| Chess Puzzles | 5% | — |
| Thematic Generalization | 53% | — |
| Surface Evolver Bench | 30.6% | — |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Gemma 4 31B IT leads
Gemma 4 31B IT: 43.2 (#81), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 8.1% |
| LMArena Math | 1465 | 1283 |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
Knowledge Gemma 4 31B IT leads
Gemma 4 31B IT: 37.9 (#151), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 75.8% | 49.1% |
| LMArena Expert | 1465 | 1245 |
| SimpleQA Verified | 10.4% | — |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| Vectara Hallucination Rate | 7.4% | — |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multimodal Not comparable
Gemma 4 31B IT: 41.6 (#34), Qwen2.5 72B Instruct: —
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Vision | 1277 | — |
| LMArena Document | 1425 | — |
Multilingual Gemma 4 31B IT leads
Gemma 4 31B IT: 53.8 (#57), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1431 | 1252 |
| LMArena Chinese | 1476 | 1272 |
| LMArena French | 1435 | 1280 |
| LMArena Russian | 1460 | 1264 |
| LMArena Spanish | 1444 | 1256 |
| LMArena German | — | 1234 |
| LMArena Japanese | — | 1180 |
| LMArena Korean | — | 1188 |
Instruction Following Gemma 4 31B IT leads
Gemma 4 31B IT: 75.5 (#61), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1433 | 1254 |
| IFEval | — | 80.6% |
Long Context Gemma 4 31B IT leads
Gemma 4 31B IT: 44.2 (#71), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1446 | 1282 |
Writing & Preference Gemma 4 31B IT leads
Gemma 4 31B IT: 60.5 (#96), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Gemma 4 31B IT | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1443 | 1269 |
| LMArena Creative Writing | 1415 | 1221 |
| LMArena Multi-Turn | 1452 | 1272 |
| EQ-Bench Creative Writing | 1368 | — |
| WildBench | — | 80.2% |
| EQ-Bench 4 | 1120 | — |
Frequently asked questions
Is Gemma 4 31B IT better than Qwen2.5 72B Instruct?
Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 31.9 on the Noometry Index.
Which is cheaper, Gemma 4 31B IT or Qwen2.5 72B Instruct?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is Gemma 4 31B IT or Qwen2.5 72B Instruct better for coding?
Gemma 4 31B IT scores higher on coding benchmarks: 42.3 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
Gemma 4 31B IT does, with 262K tokens against 131K.
How many benchmarks do Gemma 4 31B IT and Qwen2.5 72B Instruct share?
20 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and Qwen2.5 72B Instruct has 43.