LLM Hosting Pro
← All model families

Gemma

Google's lightweight open models, built from the same research as Gemini.

MultilingualSmall and fastVision

Overview

Gemma is Google DeepMind's family of lightweight open-weight models, designed to run on a single GPU or even a laptop. Later versions add vision and longer context.

Gemma uses Google's own Gemma Terms of Use, which allow commercial use but require following Google's prohibited use policy.

Models in this family

ModelReleasedParametersContext
Gemma 3 4B12 Mar 20254B128K tokens
Gemma 3 12B12 Mar 202512B128K tokens
Gemma 3 27B12 Mar 202527B128K tokens
Gemma 2 9B27 Jun 20249B8K tokens
Gemma 2 27B27 Jun 202427B8K tokens

Where to host Gemma models

All hosts →
LLM hostModelsHosting plans
Amazon AWSGemma 3 4B, Gemma 3 12B, Gemma 3 27B, Gemma 2 9B, Gemma 2 27BAmazon EC2, Amazon Bedrock, Amazon SageMaker
Google CloudGemma 3 4B, Gemma 3 12B, Gemma 3 27B, Gemma 2 9B, Gemma 2 27BGoogle Kubernetes Engine (GKE), Compute Engine, Model Garden
Microsoft AzureGemma 3 4B, Gemma 3 12B, Gemma 3 27B, Gemma 2 9B, Gemma 2 27BAzure Machine Learning, Azure Kubernetes Service (AKS), Azure GPU Virtual Machines