Google AI Edge Gallery
An open-source local AI app that runs models on-device for chat, image questions, audio transcription and private media search.
Alternative to:
Google AI Edge Gallery is a local AI app for people who want to use and compare generative models on their own devices. Model inference runs on-device, including offline chat and media tasks.
- Chat and image questions: Have multi-turn conversations or ask about camera and gallery images; supported Gemma 4 models can display reasoning output in Thinking Mode.
- Media search: EmbeddingGemma 2 supports natural-language photo searches and image similarity matching, while Video Moment Finder locates scenes by description and lets you trim and export clips.
- Audio transcription: Audio Scribe turns voice recordings into text and translates them using on-device models.
- Agent tools and device actions: Agent Skills add tools such as Wikipedia and interactive maps, while Mobile Actions uses a fine-tuned FunctionGemma 270m model for offline device controls and automated tasks.
- Model comparison: Download models through Hugging Face or load custom models, test prompts in Prompt Lab, and benchmark performance on your hardware.
The app is open source under Apache 2.0, with Android, iOS and macOS downloads. The mobile apps require Android 12 or later or iOS 17 or later. Google AI Edge and LiteRT provide the local runtime. Model downloads and connected skills such as Wikipedia require internet access; local inference doesn't send prompts, images or recordings to a model server.
Logotyper
