Gemma 3 4B

Google DeepMind

Gemma 3 4B

Developer	Google DeepMind
Release date	March 12, 2025
Family	Gemma
Generation	Gemma 3
Size class
Context window	131K tokens
Max output	16K tokens
Input	Text, image
Output	Text
Knowledge cutoff	2024-08
Weights	Open weights
Predecessor	Gemma 2 2B IT
Successor	Gemma 3n E2B IT
Website	deepmind.google
Links	Model card Hugging Face

Part of a series on

Google DeepMind

input: $0.05
output: $0.10
Context: 131Ktokens
Released: 1y ago
Size: 4Bparameters

Specifications

Model

Developer: Google DeepMind
Family: Gemma Gemma 3
Type: Language model
Released: March 12, 2025
Model ID: google/gemma-3-4b-it

Capabilities

Input → output: Text, image → Text
Native context: 128K tokens
Max output: 16K tokens
Tokenizer: Gemini
Knowledge cutoff: 2024-08

Weights

Openness: Open weights
Size: 4B
Hugging Face: google/gemma-3-4b-it

Lineage

Models built on this: Nemotron 3 Content Safetyfine-tuned from
Nemotron Content Safety Reasoning 4Bfine-tuned from

Sources:models.devOpenRouter

Metadata synced Nov 8, 2026, 1:01 PM UTC

To give you the best experience, we use cookies for personalization, insights, and performance. Learn more.