Google releases Gemma 4 12B, an encoder-free multimodal model

VentureBeat

With 11.95 billion parameters and Apache 2.0 license, it runs locally on 16GB laptops. Native audio and vision inputs flow directly into the LLM backbone without separate encoders.

GemmaGoogle

See 7 more sources

google/gemma-4-12B18 days agogoogle

google/gemma-4-12B-it-assistant17 days agogoogle

google/gemma-4-12B-it18 days agogoogle

Google’s newly released open weights model, Gemma 4 12B, supports transcription but is far from the...16 days agoArtificial Analysis

Gemma 4 12B shipped today under the label "encoder-free." A local 12b model that shows really...18 days agoKimmonismus

Introducing Gemma 4 12B: a unified, encoder-free multimodal model18 days agojohnnyApplePRNGDiscuss

Gemma 4 12B is out — interesting local LLM option for 16GB ComfyUI workflows17 days agofflluuxxuussDiscuss

VentureBeat·Carl Franzen

Jun 3, 6:49 PM