runlocal.devCheck My GPU →

Gemma 4 27B

MoEGemma

Google's 27B MoE model with only 4B active parameters per token. Near-frontier quality at a fraction of compute cost.

Provider

Google

Parameters

27B (4B active MoE)

Context

128K

Released

2026-04-08

VRAM Requirements by Quantization

MethodDisk SizeVRAM RequiredFits GPUs
Q8_028 GB30 GB3 GPUs
Q4_K_M15.5 GB17 GB7 GPUs
Q4_014.8 GB16 GB7 GPUs
Q2_K10 GB11.5 GB12 GPUs

Install with Ollama

Run in terminal:

ollama pull gemma4:27b

Minimum 11.5GB VRAM required. Install Ollama from ollama.com

Benchmark Scores

mmlu83.5%
humaneval80.1%

Scores are approximate and may vary by quantization level.

Compatible GPUs (12)

HuggingFace

google/gemma-4-27b-it

View on HF →