Google: Gemma 3 4B
Model Type
Open Weight Model
4B parameters
Recommended Use Cases
Text Generation
Try Gemma 3 4B
Gemma 3 is Google's lightweight open model family built from Gemini 2.0 research, released in March 2025, designed to run on single GPUs/TPUs with state-of-the-art performance.
Overview
Gemma 3 is a collection of lightweight, state-of-the-art open models built from the same research and technology that powers Google's Gemini 2.0 models. Available in multiple sizes (270M, 1B, 4B, 12B, 27B), these models are designed to run efficiently on a single GPU or TPU while delivering frontier-level performance.
Model Variants
| Model | Parameters | Context | Multimodal |
|---|---|---|---|
| Gemma 3 27B | 27 billion | 128K | Text + Image |
| Gemma 3 12B | 12 billion | 128K | Text + Image |
| Gemma 3 4B | 4 billion | 128K | Text + Image |
| Gemma 3 1B | 1 billion | 32K | Text only |
| Gemma 3 270M | 270 million | 32K | Text only |
Key Features
- Multimodal capabilities: 4B+ models process both text and images
- 128K context window: 16x larger than previous Gemma models (32K for 1B/270M)
- 140+ language support: Out-of-box support for 35+ languages, pretrained for 140+
- Function calling: Supports structured output and agentic workflows
- Quantized versions: Official QAT checkpoints for reduced memory with maintained accuracy
Role in Series
The Gemma family offers models optimized for different deployment scenarios:
- Gemma 3 270M: Ultra-compact for task-specific fine-tuning, extreme efficiency
- Gemma 3n E2B/E4B: Mobile-first with 2-3GB RAM, multimodal including audio
- Gemma 3 1B-27B: Cloud/desktop deployment with maximum capability (this model)
- CodeGemma: Code-specialized variant for development tasks
- MedGemma: Medical domain variant for healthcare applications