What is Gemma 4 and what sizes are available?

Gemma 4's open multimodal family: E2B (2.3B eff), E4B (4.5B eff), 26B MoE, 31B dense—all IT variants for chat/code/agents.

How do I run Gemma 4 on my GPU or edge device?

Pip transformers/accelerate, pipeline('any-to-any', model='google/gemma-4-*-it'). E2B/E4B for Pi/phone; 26B one A100; 31B two GPUs.

Gemma edges on multimodal/edge speed; Llama might win raw text scale. Test your stack—both Apache free.

🤖 AI Dev Tools

Gemma 4 drops multimodal muscle into devs' hands, offline on phones or Pis. Google's open bet challenges closed AI giants head-on.

DevTools Feed Apr 03, 2026 3 min read 28 views

Read in: Deutsch English Español Français Italiano 日本語 한국어 Português (BR) Русский Türkçe

Published by

Ship faster. Build smarter.

#Gemma 4 #Google DeepMind #edge inference #open multimodal AI

Get the best Developer Tools stories of the week in your inbox — no noise, no spam.

Originally reported by dev.to