Google AI Edge Gallery
Local AI app for Android and iOS. Run Gemma 4 on-device, ask questions about photos, transcribe audio and compare model performance.
Google AI Edge Gallery is an open-source app for people who want to try generative AI on their own phone. It runs model inference locally, so offline chat, image analysis and audio tasks don't send your inputs to a server. It supports Android and iOS and uses the Apache 2.0 license.
Gemma 4 support includes multi-turn chat and a Thinking Mode that displays the model's reasoning output. You can ask questions about pictures from your camera or photo library, identify objects and get image descriptions. Audio Scribe transcribes and translates voice recordings on-device, while Prompt Lab provides a separate space to compare prompts and adjust how a model responds.
The app also lets you explore what local models can do beyond conversation. Agent Skills add tools such as Wikipedia, interactive maps and visual summary cards, with modular skills available through URLs and community contributions. Connected skills such as Wikipedia require network access. Mobile Actions handles offline device controls and automated tasks using a fine-tuned FunctionGemma 270m model.
For people comparing models, Gallery combines Hugging Face model discovery and downloads with support for loading custom models. Its model library and benchmarks let you compare performance on the hardware you'll actually use. Google AI Edge and LiteRT provide the underlying on-device runtime. The mobile apps require Android 12 or later or iOS 17 or later.
