Skip to main content
Rase AI Intelligence Newsletter

Consumer devices Model or AI service

Google built Gemma 3n for on-device multimodal AI

Gemma 3n accepts text, images, audio and video, and generates text. Google designed the model family with phones and other edge devices in mind. [1][2]

What the model does

Google released two effective-size variants, E2B and E4B. Developers can use supported local inference tools to build and evaluate applications around either variant. [1][2]

Why it matters

Local inference gives application teams a way to process media and text on compatible devices. [1][2]

Key dates

  1. Google opens the Gemma 3n preview [2]
  2. Google announces the full Gemma 3n release [1]

Model limitations

The appropriate model and runtime depend on the device’s resources and the application’s inputs. [1][2]

Sources and verification

Catalog evidence checked How we verify records

  1. Primary source Google Gemma 3n developer guide
  2. Primary source Google announces the Gemma 3n preview

Related profiles

Consumer devices

Samsung Ballie

Samsung Electronics

Explores mobile home-assistant interactions, including projected information, through a robot concept.

Medium confidenceEvidence checked

Consumer devices

Alibaba Quark AI Glasses

Alibaba (Quark)

Lets users interact with Qwen through supported audio and visual inputs, with features varying by model.

High confidenceEvidence checked