Model comparison
Gemma 3 27B vs GPT-4o
Gemma 3 27B is the stronger model overall, scoring 30.5 to 28.5 on the Noometry Index.
Last verified . 39 shared benchmarks.
GoogleSummary
- They share 39 benchmarks with published results for both. Gemma 3 27B scores higher in 5 categories and GPT-4o in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 3 27B leads 24.4 to 10.2.
- The biggest single-benchmark swing is Aider Polyglot: 4.9% for Gemma 3 27B and 45.3% for GPT-4o.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Gemma 3 27B accepts more context: 131K tokens versus 128K.
- Gemma 3 27B has downloadable open weights; the other is API-only.
Side by side
| Gemma 3 27B | GPT-4o | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 30.5 | 28.5 |
| Released | 2025-03-11 | 2024-05-13 |
| Weights | Open | Proprietary |
| Context window | 131K | 128K |
| Max output | 8K | 16K |
| Input $ / M tokens | $0.08 | $2.50 |
| Output $ / M tokens | $0.16 | $10 |
| Results tracked | 43 | 72 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4o leads
Gemma 3 27B: 22.1 (#329), GPT-4o: 25.1 (#323)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| Aider Polyglot | 4.9% | 45.3% |
| LiveBench Coding | 39.9% | 51.4% |
| LMArena Coding | 1322 | 1297 |
| SWE-bench Verified | — | 31% |
| SWE-bench Verified (bash only) | — | 21.6% |
| SciCode | 21.2% | — |
| GSO | — | 0% |
| WeirdML | — | 25.1% |
| BigCodeBench Instruct | — | 51.1% |
| BigCodeBench Complete | — | 61.1% |
| CadEval | — | 26% |
| HumanEval+ | — | 87.2% |
| MBPP+ | — | 72.2% |
Agentic & Tool Use Gemma 3 27B leads
Gemma 3 27B: 24.9 (#105), GPT-4o: 20.8 (#133)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| Berkeley Function Calling Leaderboard | 29.5% | — |
| GDPval | — | 9.9% |
| TheAgentCompany | — | 8.6% |
| Cybench | — | 12.5% |
| BALROG | — | 32.3% |
| LMArena Search | — | 1006 |
| METR Time Horizons | — | 40.8% |
Reasoning Gemma 3 27B leads
Gemma 3 27B: 16.3 (#297), GPT-4o: 9.4 (#338)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| CritPt | 0% | 0% |
| Chess Puzzles | 0% | 13% |
| LiveBench Reasoning | 43.8% | 55.8% |
| LMArena Hard Prompts | 1340 | 1281 |
| DTBench | 52.5% | 64.5% |
| LiveBench Data Analysis | 51.5% | 60.9% |
| LMCA | 12.3% | 16.6% |
| Epoch Capabilities Index | 130.04 | 128.97 |
| LiveBench | 50% | 55.3% |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 17.8% |
| Kagi LLM Benchmark | 40.4% | — |
| ARC-AGI-1 | — | 4.5% |
| EnigmaEval | — | 0.8% |
| ForecastBench | — | 57.7 |
Math Gemma 3 27B leads
Gemma 3 27B: 24.4 (#261), GPT-4o: 10.2 (#303)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 22.5% | 6.4% |
| LiveBench Math | 55.4% | 49.5% |
| LMArena Math | 1312 | 1285 |
| MATH Level 5 | 74% | 53.3% |
| FrontierMath (Tiers 1-3) | — | 0.4% |
| Omni-MATH | — | 29.3% |
| FrontierMath (Feb 2025 set) | — | 0.3% |
Knowledge GPT-4o leads
Gemma 3 27B: 25.4 (#256), GPT-4o: 28.8 (#235)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| GPQA Diamond | 47.7% | 49.2% |
| Confabulations | 40.3% | 15.3% |
| Vectara Hallucination Rate | 7.4% | 9.6% |
| LMArena Expert | 1303 | 1250 |
| Humanity's Last Exam | — | 2.7% |
| SimpleQA Verified | — | 26% |
| MMLU-Pro | — | 71.3% |
| GPQA (HELM) | — | 52% |
| MMLU | — | 88.1% |
Multimodal GPT-4o leads
Gemma 3 27B: 32.0 (#94), GPT-4o: 33.9 (#85)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| LMArena Vision | 1165 | 1137 |
| GeoBench | 52% | 71% |
| Video-MME | — | 71.9% |
| VPCT | — | 40% |
| ScienceQA | — | 88.5% |
Multilingual Gemma 3 27B leads
Gemma 3 27B: 47.0 (#150), GPT-4o: 43.4 (#181)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| LMArena Non-English | 1334 | 1283 |
| LMArena Chinese | 1347 | 1277 |
| LMArena French | 1368 | 1303 |
| LMArena German | 1362 | 1282 |
| LMArena Japanese | 1287 | 1257 |
| LMArena Korean | 1308 | 1234 |
| LMArena Russian | 1349 | 1286 |
| LMArena Spanish | 1348 | 1292 |
Instruction Following Gemma 3 27B leads
Gemma 3 27B: 70.8 (#155), GPT-4o: 66.8 (#202)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| LiveBench Instruction Following | 74.9% | 68.6% |
| LMArena Instruction Following | 1321 | 1278 |
| IFEval | — | 81.7% |
Long Context GPT-4o leads
Gemma 3 27B: 27.6 (#288), GPT-4o: 39.4 (#174)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| Fiction.LiveBench | 33.3% | 66.7% |
| LMArena Longer Query | 1333 | 1289 |
Writing & Preference Too close to call
Gemma 3 27B: 52.7 (#161), GPT-4o: 52.8 (#160)
| Benchmark | Gemma 3 27B | GPT-4o |
|---|---|---|
| LMArena Text | 1358 | 1300 |
| LMArena Creative Writing | 1346 | 1291 |
| Short-Story Creative Writing | 79.9% | 81.8% |
| LMArena Multi-Turn | 1344 | 1302 |
| LiveBench Language | 34.6% | 47.6% |
| EQ-Bench Creative Writing | 1266 | — |
| WildBench | — | 82.8% |
Frequently asked questions
Is Gemma 3 27B better than GPT-4o?
Gemma 3 27B is the stronger model overall, scoring 30.5 to 28.5 on the Noometry Index.
Which is cheaper, Gemma 3 27B or GPT-4o?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; GPT-4o lists at $2.50 and $10.
Is Gemma 3 27B or GPT-4o better for coding?
GPT-4o scores higher on coding benchmarks: 25.1 versus 22.1 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 27B does, with 131K tokens against 128K.
How many benchmarks do Gemma 3 27B and GPT-4o share?
39 benchmarks have published results for both models. Gemma 3 27B has 43 scored results on Noometry and GPT-4o has 72.