# Gemma 3 12B vs GPT-4

> Gemma 3 12B is the stronger model overall, scoring 31.8 to 28.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-3-12b-vs-gpt-4
- Last updated: 2026-10-10
- Shared benchmarks: 19

## Summary

- They share 19 benchmarks with published results for both. Gemma 3 12B scores higher in 7 categories and GPT-4 in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemma 3 12B leads 47.7 to 35.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 16.7% for Gemma 3 12B and 1.1% for GPT-4.
- Gemma 3 12B is cheaper at $0.05 / $0.15 per million input/output tokens, against $30 / $60 for GPT-4.
- Gemma 3 12B accepts more context: 131K tokens versus 8K.
- Gemma 3 12B has downloadable open weights; the other is API-only.

## Snapshot

| | Gemma 3 12B | GPT-4 |
|---|---|---|
| Provider | Google | OpenAI |
| Noometry Index | 31.8 | 28.9 |
| Rank | 255 | 308 |
| Context | 131K | 8K |
| Input $/M | $0.05 | $30 |
| Output $/M | $0.15 | $60 |
| Weights | Open | Proprietary |

## Coding

- Gemma 3 12B: 31.2 (#276)
- GPT-4: 31.2 (#277)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| LMArena Coding | 1281 | 1254 |
| SciCode | 17.4% | — |
| WeirdML | — | 12.4% |
| BigCodeBench Instruct | — | 46% |
| BigCodeBench Complete | — | 57.2% |
| HumanEval+ | — | 79.3% |

## Agentic & Tool Use

- Gemma 3 12B: 25.3 (#101)
- GPT-4: —

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 30.4% | — |
| METR Time Horizons | — | 36.1% |

## Reasoning

- Gemma 3 12B: 15.2 (#307)
- GPT-4: 17.3 (#285)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| Chess Puzzles | 0% | 4% |
| LMArena Hard Prompts | 1309 | 1241 |
| DTBench | 48.8% | 62.7% |
| LMCA | 4.5% | 17.1% |
| Epoch Capabilities Index | 123.49 | 125.88 |
| CritPt | 0% | — |
| Mystery Game Puzzles | — | 12% |
| BIG-Bench Hard | — | 75.1% |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |

## Math

- Gemma 3 12B: 21.0 (#273)
- GPT-4: 10.1 (#304)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 16.7% | 1.1% |
| LMArena Math | 1307 | 1269 |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |

## Knowledge

- Gemma 3 12B: 26.5 (#250)
- GPT-4: 18.4 (#277)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| GPQA Diamond | 39.5% | 35.7% |
| LMArena Expert | 1248 | 1211 |
| Vectara Hallucination Rate | 4.4% | — |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |

## Multimodal

- Gemma 3 12B: —
- GPT-4: —

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| MindCube | 46.7% | — |

## Multilingual

- Gemma 3 12B: 45.9 (#160)
- GPT-4: 40.8 (#210)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| LMArena Non-English | 1318 | 1246 |
| LMArena German | 1370 | 1251 |
| LMArena Russian | 1335 | 1251 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1282 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Spanish | — | 1261 |

## Instruction Following

- Gemma 3 12B: 68.8 (#179)
- GPT-4: 65.5 (#217)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| LMArena Instruction Following | 1299 | 1241 |

## Long Context

- Gemma 3 12B: 40.0 (#157)
- GPT-4: 37.7 (#207)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| LMArena Longer Query | 1317 | 1244 |

## Writing & Preference

- Gemma 3 12B: 47.7 (#205)
- GPT-4: 35.0 (#263)

| Benchmark | Gemma 3 12B | GPT-4 |
|---|---|---|
| LMArena Text | 1334 | 1263 |
| LMArena Creative Writing | 1331 | 1244 |
| EQ-Bench Creative Writing | 1126 | 752 |
| LMArena Multi-Turn | 1334 | 1257 |

## FAQ

### Is Gemma 3 12B better than GPT-4?

Gemma 3 12B is the stronger model overall, scoring 31.8 to 28.9 on the Noometry Index.

### Which is cheaper, Gemma 3 12B or GPT-4?

Gemma 3 12B is cheaper. It lists at $0.05 per million input tokens and $0.15 per million output tokens; GPT-4 lists at $30 and $60.

### Is Gemma 3 12B or GPT-4 better for coding?

They score almost the same on coding (31.2 vs 31.2); test both on your own repository before choosing.

### Which has the bigger context window?

Gemma 3 12B does, with 131K tokens against 8K.

### How many benchmarks do Gemma 3 12B and GPT-4 share?

19 benchmarks have published results for both models. Gemma 3 12B has 24 scored results on Noometry and GPT-4 has 38.