Gemma 4 31B

Google DeepMind

Gemma 4 31B

Developer	Google DeepMind
Release date	April 2, 2026
Family	Gemma
Generation	Gemma 4
Size class
Architecture	gemma4
Context window	262K tokens
Max output	16K tokens
Input	Image, text, video
Output	Text
Knowledge cutoff	2025-01
Weights	Open weights
Predecessor	Gemma 3n E2B IT
Website	deepmind.google
Links	Model card Docs Hugging Face

Part of a series on

Google DeepMind

input: $0.20
output: $0.40
Context: 262Ktokens
Released: 7mo ago
Size: 32.7Bparameters

Specifications

Model

Developer: Google DeepMind
Family: Gemma Gemma 4
Type: Language model
Released: April 2, 2026
Model ID: google/gemma-4-31b-it

Capabilities

Input → output: Image, text, video → Text
Native context: 262K tokens
Max output: 16K tokens
Tokenizer: Gemma
Knowledge cutoff: 2025-01

Weights

Openness: Open weights
Size: 32.7B
Architecture: gemma4
Families: gemma4
Hugging Face: google/gemma-4-31B-it

Sources:models.devOpenRouter

Metadata synced Nov 5, 2026, 10:03 AM UTC

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