AI BriefingGoogle AI Edge Gallery - Open-Source Fully Offline LLM Gallery App
Key point
Google has released Google AI Edge Gallery, an open-source app for running on-device generative AI.
Details
Google has released Google AI Edge Gallery, an open-source app for iOS/Android that runs LLMs in a fully offline and private environment without an internet connection.
The app officially supports the Gemma 4 family, allowing you to test advanced reasoning, logic, and creative capabilities on-device without sending data to a server. Key features include:
- Agent Skills: Support for modular skills that extend the LLM's capabilities using external tools such as Wikipedia search and interactive maps
- Thinking Mode: Visually view the model's step-by-step reasoning process
- Multimodal & Audio: Support for camera-based object recognition (Ask Image) and real-time voice transcription/translation (Audio Scribe)
- Mobile Actions: A feature that automates device control based on the FunctionGemma model
- Prompt Lab: A dedicated workspace for finely tuning parameters such as temperature and top-k
By integrating the lightweight LiteRT runtime with Hugging Face, the app supports model discovery and optimization, and can be installed and used on Android 12+ and iOS 17+ environments.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.