Model comparison
Gemma 4 31B IT vs GPT-3.5-turbo
Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 23.2 on the Noometry Index.
Last verified . 22 shared benchmarks.
GoogleSummary
- They share 22 benchmarks with published results for both. Gemma 4 31B IT scores higher in 8 categories and GPT-3.5-turbo 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 6.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 73.3% for Gemma 4 31B IT and 2.2% for GPT-3.5-turbo.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $0.50 / $1.50 for GPT-3.5-turbo.
- Gemma 4 31B IT accepts more context: 262K tokens versus 16K.
- Gemma 4 31B IT has downloadable open weights; the other is API-only.
Side by side
| Gemma 4 31B IT | GPT-3.5-turbo | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 43.5 | 23.2 |
| Released | 2026-04-02 | 2023-03-01 |
| Weights | Open | Proprietary |
| Context window | 262K | 16K |
| Max output | 33K | 4K |
| Input $ / M tokens | $0.09 | $0.50 |
| Output $ / M tokens | $0.34 | $1.50 |
| Results tracked | 35 | 44 |
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), GPT-3.5-turbo: 23.9 (#331)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| WeirdML | 52.3% | 3.5% |
| LMArena Coding | 1459 | 1136 |
| LMArena WebDev | 1366 | — |
| SciCode | 43.4% | — |
| BigCodeBench Instruct | — | 39.1% |
| BigCodeBench Complete | — | 50.6% |
| ALE-Bench | 925.5 | — |
| HumanEval+ | — | 70.7% |
| MBPP+ | — | 69.7% |
Agentic & Tool Use Not comparable
Gemma 4 31B IT: —, GPT-3.5-turbo: —
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| METR Time Horizons | — | 21.5% |
Reasoning Gemma 4 31B IT leads
Gemma 4 31B IT: 27.2 (#122), GPT-3.5-turbo: 13.8 (#332)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| Chess Puzzles | 5% | 0% |
| LMArena Hard Prompts | 1448 | 1108 |
| DTBench | 82.7% | 48.5% |
| LMCA | 39.3% | 9.7% |
| Epoch Capabilities Index | 142.74 | 118.55 |
| Kagi LLM Benchmark | 63.5% | — |
| NYT Connections (extended) | 70.6% | — |
| CritPt | 1.4% | — |
| Thematic Generalization | 53% | — |
| Mystery Game Puzzles | — | 3% |
| Surface Evolver Bench | 30.6% | — |
| Adversarial NLI | — | 58.1% |
| BIG-Bench Hard | — | 61.6% |
| CommonsenseQA 2.0 | — | 57% |
| ForecastBench | — | 50.4 |
| WinoGrande | — | 81.6% |
Math Gemma 4 31B IT leads
Gemma 4 31B IT: 43.2 (#81), GPT-3.5-turbo: 6.3 (#327)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 2.2% |
| LMArena Math | 1465 | 1142 |
| FrontierMath (Tiers 1-3) | — | 0% |
| MATH Level 5 | — | 15.9% |
| GSM8K | — | 57.8% |
Knowledge Gemma 4 31B IT leads
Gemma 4 31B IT: 37.9 (#151), GPT-3.5-turbo: 10.0 (#303)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| GPQA Diamond | 75.8% | 28% |
| LMArena Expert | 1465 | 1070 |
| SimpleQA Verified | 10.4% | — |
| Vectara Hallucination Rate | 7.4% | — |
| ARC (AI2) Challenge | — | 87.4% |
| BoolQ | — | 87% |
| MMLU | — | 71.4% |
| OpenBookQA | — | 86% |
| TriviaQA | — | 85.8% |
Multimodal Not comparable
Gemma 4 31B IT: 41.6 (#34), GPT-3.5-turbo: —
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| LMArena Vision | 1277 | — |
| LMArena Document | 1425 | — |
Multilingual Gemma 4 31B IT leads
Gemma 4 31B IT: 53.8 (#57), GPT-3.5-turbo: 31.5 (#258)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| LMArena Non-English | 1431 | 1108 |
| LMArena Chinese | 1476 | 1075 |
| LMArena French | 1435 | 1118 |
| LMArena Russian | 1460 | 1123 |
| LMArena Spanish | 1444 | 1121 |
| LMArena German | — | 1090 |
| LMArena Japanese | — | 1043 |
| LMArena Korean | — | 1019 |
Instruction Following Gemma 4 31B IT leads
Gemma 4 31B IT: 75.5 (#61), GPT-3.5-turbo: 57.9 (#262)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| LMArena Instruction Following | 1433 | 1119 |
Long Context Gemma 4 31B IT leads
Gemma 4 31B IT: 44.2 (#71), GPT-3.5-turbo: 34.0 (#254)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| LMArena Longer Query | 1446 | 1121 |
Writing & Preference Gemma 4 31B IT leads
Gemma 4 31B IT: 60.5 (#96), GPT-3.5-turbo: 25.3 (#305)
| Benchmark | Gemma 4 31B IT | GPT-3.5-turbo |
|---|---|---|
| LMArena Text | 1443 | 1125 |
| LMArena Creative Writing | 1415 | 1092 |
| EQ-Bench Creative Writing | 1368 | 451 |
| LMArena Multi-Turn | 1452 | 1117 |
| EQ-Bench 4 | 1120 | — |
Frequently asked questions
Is Gemma 4 31B IT better than GPT-3.5-turbo?
Gemma 4 31B IT is the stronger model overall, scoring 43.5 to 23.2 on the Noometry Index.
Which is cheaper, Gemma 4 31B IT or GPT-3.5-turbo?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; GPT-3.5-turbo lists at $0.50 and $1.50.
Is Gemma 4 31B IT or GPT-3.5-turbo better for coding?
Gemma 4 31B IT scores higher on coding benchmarks: 42.3 versus 23.9 in the Noometry coding category.
Which has the bigger context window?
Gemma 4 31B IT does, with 262K tokens against 16K.
How many benchmarks do Gemma 4 31B IT and GPT-3.5-turbo share?
22 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and GPT-3.5-turbo has 44.