Model comparison
Gemma 3 27B vs GPT-3.5-turbo
Gemma 3 27B is the stronger model overall, scoring 30.8 to 23.2 on the Noometry Index.
Last verified . 25 shared benchmarks.
GoogleSummary
- They share 25 benchmarks with published results for both. Gemma 3 27B scores higher in 6 categories and GPT-3.5-turbo in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemma 3 27B leads 52.5 to 25.3.
- The biggest single-benchmark swing is MATH Level 5: 74% for Gemma 3 27B and 15.9% for GPT-3.5-turbo.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $0.50 / $1.50 for GPT-3.5-turbo.
- Gemma 3 27B accepts more context: 131K tokens versus 16K.
- Gemma 3 27B has downloadable open weights; the other is API-only.
Side by side
| Gemma 3 27B | GPT-3.5-turbo | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 30.8 | 23.2 |
| Released | 2025-03-11 | 2023-03-01 |
| Weights | Open | Proprietary |
| Context window | 131K | 16K |
| Max output | 8K | 4K |
| Input $ / M tokens | $0.08 | $0.50 |
| Output $ / M tokens | $0.16 | $1.50 |
| Results tracked | 43 | 44 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-3.5-turbo leads
Gemma 3 27B: 22.5 (#334), GPT-3.5-turbo: 23.9 (#331)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Coding | 1322 | 1136 |
| Aider Polyglot | 4.9% | — |
| SciCode | 21.2% | — |
| WeirdML | — | 3.5% |
| BigCodeBench Instruct | — | 39.1% |
| LiveBench Coding | 39.9% | — |
| BigCodeBench Complete | — | 50.6% |
| HumanEval+ | — | 70.7% |
| MBPP+ | — | 69.7% |
Agentic & Tool Use Not comparable
Gemma 3 27B: 25.1 (#110), GPT-3.5-turbo: —
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| Berkeley Function Calling Leaderboard | 29.5% | — |
| METR Time Horizons | — | 21.5% |
Reasoning Gemma 3 27B leads
Gemma 3 27B: 16.7 (#301), GPT-3.5-turbo: 13.8 (#332)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| Chess Puzzles | 0% | 0% |
| LMArena Hard Prompts | 1340 | 1108 |
| DTBench | 52.5% | 48.5% |
| LMCA | 12.3% | 9.7% |
| Epoch Capabilities Index | 130.04 | 118.55 |
| Kagi LLM Benchmark | 40.4% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 43.8% | — |
| Mystery Game Puzzles | — | 3% |
| LiveBench Data Analysis | 51.5% | — |
| Adversarial NLI | — | 58.1% |
| BIG-Bench Hard | — | 61.6% |
| CommonsenseQA 2.0 | — | 57% |
| ForecastBench | — | 50.4 |
| LiveBench | 50% | — |
| WinoGrande | — | 81.6% |
Math Gemma 3 27B leads
Gemma 3 27B: 25.9 (#265), GPT-3.5-turbo: 6.3 (#327)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 22.5% | 2.2% |
| LMArena Math | 1312 | 1142 |
| MATH Level 5 | 74% | 15.9% |
| FrontierMath (Tiers 1-3) | — | 0% |
| LiveBench Math | 55.4% | — |
| GSM8K | — | 57.8% |
Knowledge Gemma 3 27B leads
Gemma 3 27B: 25.5 (#261), GPT-3.5-turbo: 10.0 (#303)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| GPQA Diamond | 47.7% | 28% |
| LMArena Expert | 1304 | 1070 |
| Confabulations | 40.3% | — |
| Vectara Hallucination Rate | 7.4% | — |
| ARC (AI2) Challenge | — | 87.4% |
| BoolQ | — | 87% |
| MMLU | — | 71.4% |
| OpenBookQA | — | 86% |
| TriviaQA | — | 85.8% |
Multimodal Not comparable
Gemma 3 27B: 32.6 (#100), GPT-3.5-turbo: —
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Vision | 1164 | — |
| GeoBench | 52% | — |
Multilingual Gemma 3 27B leads
Gemma 3 27B: 46.9 (#155), GPT-3.5-turbo: 31.5 (#258)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Non-English | 1334 | 1108 |
| LMArena Chinese | 1346 | 1075 |
| LMArena French | 1368 | 1118 |
| LMArena German | 1362 | 1090 |
| LMArena Japanese | 1287 | 1043 |
| LMArena Korean | 1308 | 1019 |
| LMArena Russian | 1349 | 1123 |
| LMArena Spanish | 1349 | 1121 |
Instruction Following Gemma 3 27B leads
Gemma 3 27B: 70.6 (#160), GPT-3.5-turbo: 57.9 (#262)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Instruction Following | 1321 | 1119 |
| LiveBench Instruction Following | 74.9% | — |
Long Context GPT-3.5-turbo leads
Gemma 3 27B: 27.6 (#293), GPT-3.5-turbo: 34.0 (#254)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Longer Query | 1333 | 1121 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Gemma 3 27B leads
Gemma 3 27B: 52.5 (#168), GPT-3.5-turbo: 25.3 (#305)
| Benchmark | Gemma 3 27B | GPT-3.5-turbo |
|---|---|---|
| LMArena Text | 1358 | 1125 |
| LMArena Creative Writing | 1346 | 1092 |
| EQ-Bench Creative Writing | 1266 | 451 |
| LMArena Multi-Turn | 1345 | 1117 |
| Short-Story Creative Writing | 79.9% | — |
| LiveBench Language | 34.6% | — |
Frequently asked questions
Is Gemma 3 27B better than GPT-3.5-turbo?
Gemma 3 27B is the stronger model overall, scoring 30.8 to 23.2 on the Noometry Index.
Which is cheaper, Gemma 3 27B or GPT-3.5-turbo?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; GPT-3.5-turbo lists at $0.50 and $1.50.
Is Gemma 3 27B or GPT-3.5-turbo better for coding?
GPT-3.5-turbo scores higher on coding benchmarks: 23.9 versus 22.5 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 27B does, with 131K tokens against 16K.
How many benchmarks do Gemma 3 27B and GPT-3.5-turbo share?
25 benchmarks have published results for both models. Gemma 3 27B has 43 scored results on Noometry and GPT-3.5-turbo has 44.