# Gemma 3 4B vs Llama 3-8B

> Gemma 3 4B is the stronger model overall, scoring 27.9 to 25.3 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gemma-3-4b-vs-llama-3-8b
- Last updated: 2026-10-10
- Shared benchmarks: 17

## Summary

- They share 17 benchmarks with published results for both. Gemma 3 4B scores higher in 7 categories and Llama 3-8B in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in multilingual, where Gemma 3 4B leads 42.7 to 31.0.
- The biggest single-benchmark swing is DTBench: 50.9% for Gemma 3 4B and 43.9% for Llama 3-8B.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.14 / $0.14 for Llama 3-8B.
- Gemma 3 4B accepts more context: 131K tokens versus 8K.

## Snapshot

| | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| Provider | Google | Meta |
| Noometry Index | 27.9 | 25.3 |
| Rank | 320 | 338 |
| Context | 131K | 8K |
| Input $/M | $0.04 | $0.14 |
| Output $/M | $0.08 | $0.14 |
| Weights | Open | Open |

## Coding

- Gemma 3 4B: 35.3 (#214)
- Llama 3-8B: 30.5 (#285)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1230 | 1152 |
| BigCodeBench Instruct | — | 31.9% |
| BigCodeBench Complete | — | 36.9% |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |

## Agentic & Tool Use

- Gemma 3 4B: 20.7 (#134)
- Llama 3-8B: —

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 19.6% | — |

## Reasoning

- Gemma 3 4B: 12.7 (#330)
- Llama 3-8B: 13.8 (#321)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| Chess Puzzles | 0% | 0% |
| LMArena Hard Prompts | 1253 | 1133 |
| DTBench | 50.9% | 43.9% |
| Epoch Capabilities Index | 116.02 | 116.33 |
| Kagi LLM Benchmark | 25.2% | — |
| LMCA | 2.8% | — |
| Adversarial NLI | — | 57.3% |
| ForecastBench | — | 58.6 |
| WinoGrande | — | 75.7% |

## Math

- Gemma 3 4B: 15.8 (#286)
- Llama 3-8B: 8.3 (#317)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 7.5% | 1.9% |
| LMArena Math | 1239 | 1151 |
| MATH Level 5 | — | 6.1% |

## Knowledge

- Gemma 3 4B: 11.8 (#294)
- Llama 3-8B: 7.8 (#303)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| GPQA Diamond | 23.2% | 26.1% |
| LMArena Expert | 1223 | 1113 |
| Vectara Hallucination Rate | 6.4% | — |
| ARC (AI2) Challenge | — | 82.8% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |

## Multilingual

- Gemma 3 4B: 42.7 (#189)
- Llama 3-8B: 31.0 (#256)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1273 | 1098 |
| LMArena German | 1281 | 1104 |
| LMArena Russian | 1294 | 1109 |
| LMArena Chinese | — | 1076 |
| LMArena French | — | 1158 |
| LMArena Japanese | — | 967 |
| LMArena Korean | — | 1003 |
| LMArena Spanish | — | 1172 |

## Instruction Following

- Gemma 3 4B: 65.4 (#219)
- Llama 3-8B: 58.7 (#255)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1239 | 1127 |

## Long Context

- Gemma 3 4B: 38.6 (#189)
- Llama 3-8B: 34.2 (#246)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1274 | 1128 |

## Writing & Preference

- Gemma 3 4B: 42.2 (#234)
- Llama 3-8B: 37.7 (#251)

| Benchmark | Gemma 3 4B | Llama 3-8B |
|---|---|---|
| LMArena Text | 1291 | 1166 |
| LMArena Creative Writing | 1270 | 1150 |
| LMArena Multi-Turn | 1255 | 1152 |
| EQ-Bench Creative Writing | 1068 | — |

## FAQ

### Is Gemma 3 4B better than Llama 3-8B?

Gemma 3 4B is the stronger model overall, scoring 27.9 to 25.3 on the Noometry Index.

### Which is cheaper, Gemma 3 4B or Llama 3-8B?

Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; Llama 3-8B lists at $0.14 and $0.14.

### Is Gemma 3 4B or Llama 3-8B better for coding?

Gemma 3 4B scores higher on coding benchmarks: 35.3 versus 30.5 in the Noometry coding category.

### Which has the bigger context window?

Gemma 3 4B does, with 131K tokens against 8K.

### How many benchmarks do Gemma 3 4B and Llama 3-8B share?

17 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and Llama 3-8B has 34.