Model comparison
Gemma 3 27B vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 30.8 on the Noometry Index. Gemma 3 27B costs 34× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
GoogleSummary
- They share 35 benchmarks with published results for both. Gemma 3 27B scores higher in 0 categories and GPT-5 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 27.6.
- The biggest single-benchmark swing is Aider Polyglot: 4.9% for Gemma 3 27B and 88% for GPT-5.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 131K.
- Gemma 3 27B has downloadable open weights; the other is API-only.
Side by side
| Gemma 3 27B | GPT-5 | |
|---|---|---|
| Provider | OpenAI | |
| Noometry Index | 30.8 | 50.9 |
| Released | 2025-03-11 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 131K | 400K |
| Max output | 8K | 128K |
| Input $ / M tokens | $0.08 | $1.25 |
| Output $ / M tokens | $0.16 | $10 |
| Results tracked | 43 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
Gemma 3 27B: 22.5 (#334), GPT-5: 50.3 (#47)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| Aider Polyglot | 4.9% | 88% |
| SciCode | 21.2% | 42.9% |
| LMArena Coding | 1322 | 1436 |
| SWE-bench Verified | — | 73.6% |
| SWE-bench Verified (bash only) | — | 65% |
| LMArena WebDev | — | 1418 |
| GSO | — | 6.9% |
| WeirdML | — | 60.7% |
| LiveBench Coding | 39.9% | — |
| ALE-Bench | — | 1,162 |
| AlgoTune | — | 1.67 |
Agentic & Tool Use GPT-5 leads
Gemma 3 27B: 25.1 (#110), GPT-5: 33.1 (#56)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| Terminal-Bench | — | 49.6% |
| Berkeley Function Calling Leaderboard | 29.5% | — |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| DeepResearch Bench | — | 49.6% |
| BALROG | — | 32.8% |
| LMArena Search | — | 1133 |
| METR Time Horizons | — | 69.6% |
Reasoning GPT-5 leads
Gemma 3 27B: 16.7 (#301), GPT-5: 38.3 (#64)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| Kagi LLM Benchmark | 40.4% | 72.7% |
| CritPt | 0% | 12.6% |
| Chess Puzzles | 0% | 37% |
| LMArena Hard Prompts | 1340 | 1416 |
| DTBench | 52.5% | 90.7% |
| LMCA | 12.3% | 40% |
| Epoch Capabilities Index | 130.04 | 150 |
| ARC-AGI-2 | — | 9.9% |
| SimpleBench | — | 56.7% |
| ARC-AGI-1 | — | 65.7% |
| EnigmaEval | — | 10.5% |
| EBR-Bench | — | 12.7% |
| LiveBench Reasoning | 43.8% | — |
| Mystery Game Puzzles | — | 23% |
| LiveBench Data Analysis | 51.5% | — |
| ForecastBench | — | 61.4 |
| LiveBench | 50% | — |
Math GPT-5 leads
Gemma 3 27B: 25.9 (#265), GPT-5: 55.0 (#44)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 22.5% | 91.4% |
| LMArena Math | 1312 | 1407 |
| MATH Level 5 | 74% | 98.1% |
| FrontierMath (Tiers 1-3) | — | 55.4% |
| FrontierMath Tier 4 | — | 22% |
| ProofBench | — | 18% |
| Omni-MATH | — | 64.7% |
| LiveBench Math | 55.4% | — |
| FrontierMath (Feb 2025 set) | — | 32.4% |
| FrontierMath Tier 4 (v1) | — | 12.5% |
Knowledge GPT-5 leads
Gemma 3 27B: 25.5 (#261), GPT-5: 56.6 (#43)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| GPQA Diamond | 47.7% | 86.2% |
| Confabulations | 40.3% | 10.3% |
| Vectara Hallucination Rate | 7.4% | 14.7% |
| LMArena Expert | 1304 | 1419 |
| Humanity's Last Exam | — | 25.3% |
| SimpleQA Verified | — | 50.1% |
| MMLU-Pro | — | 86.3% |
| GPQA (HELM) | — | 79.2% |
Multimodal GPT-5 leads
Gemma 3 27B: 32.6 (#100), GPT-5: 46.8 (#13)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| LMArena Vision | 1164 | 1232 |
| GeoBench | 52% | 81% |
| VPCT | — | 66% |
Multilingual GPT-5 leads
Gemma 3 27B: 46.9 (#155), GPT-5: 51.4 (#110)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| LMArena Non-English | 1334 | 1397 |
| LMArena Chinese | 1346 | 1422 |
| LMArena French | 1368 | 1410 |
| LMArena German | 1362 | 1416 |
| LMArena Japanese | 1287 | 1409 |
| LMArena Korean | 1308 | 1360 |
| LMArena Russian | 1349 | 1406 |
| LMArena Spanish | 1349 | 1399 |
Instruction Following GPT-5 leads
Gemma 3 27B: 70.6 (#160), GPT-5: 73.8 (#113)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| LMArena Instruction Following | 1321 | 1388 |
| LiveBench Instruction Following | 74.9% | — |
| IFEval | — | 87.5% |
Long Context GPT-5 leads
Gemma 3 27B: 27.6 (#293), GPT-5: 69.5 (#2)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| Fiction.LiveBench | 33.3% | 97.2% |
| LMArena Longer Query | 1333 | 1399 |
Writing & Preference GPT-5 leads
Gemma 3 27B: 52.5 (#168), GPT-5: 63.4 (#65)
| Benchmark | Gemma 3 27B | GPT-5 |
|---|---|---|
| LMArena Text | 1358 | 1406 |
| LMArena Creative Writing | 1346 | 1365 |
| Short-Story Creative Writing | 79.9% | 86% |
| EQ-Bench Creative Writing | 1266 | 1627 |
| LMArena Multi-Turn | 1345 | 1426 |
| WildBench | — | 85.7% |
| LiveBench Language | 34.6% | — |
Frequently asked questions
Is Gemma 3 27B better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 30.8 on the Noometry Index. Gemma 3 27B costs 34× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 27B or GPT-5?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; GPT-5 lists at $1.25 and $10.
Is Gemma 3 27B or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 22.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 131K.
How many benchmarks do Gemma 3 27B and GPT-5 share?
35 benchmarks have published results for both models. Gemma 3 27B has 43 scored results on Noometry and GPT-5 has 69.