google-ai-edge/gallery

Google AI Edge Gallery is a mobile app that lets you run large language models, AI that can hold conversations, answer questions, and reason through problems, entirely on your phone, with no internet connection required. The models run directly on your device's hardware, so nothing leaves your phone. The problem it solves: most AI assistants send your questions to remote servers, raising privacy concerns and requiring a connection. This app lets you experience powerful AI capabilities offline and privately, on your own device. How it works: you download the app, choose one of the supported open-source language models (such as Google's Gemma 4 family), and the model is downloaded and stored on your device. From there, all processing happens locally. The app includes several built-in use cases: AI Chat (multi-turn conversations with optional step-by-step reasoning display), Ask Image (analyze photos from your camera or gallery), Audio Scribe (real-time voice transcription and translation), Prompt Lab (a testing workspace for trying different inputs), and Mobile Actions (offline device automation). You can also load custom models. You would use this app if you want to experiment with on-device AI privately, evaluate model performance on your specific phone, or build understanding of what current AI models can do locally without a server. The tech stack is Kotlin (Android) and Swift (iOS), using Google AI Edge and LiteRT for on-device model inference.