Box

•
1.0k
•

Offers a private, on-device AI suite for Android, featuring local LLM chat, image generation, voice mode, vision AI, and document analysis with advanced privacy and security.

Box transforms your Android device into a powerful, private AI hub, building upon the Google AI Edge Gallery with significant enhancements. It enables completely on-device AI capabilities, including local large language model (LLM) chat using llama.cpp and GGUF models, speech-to-speech voice interaction, vision AI, and generative image creation via stable-diffusion.cpp. Designed with privacy as a core principle, Box features a biometric lock, encrypted chat history, and the option for a hard offline mode, ensuring your data remains entirely on your device. It leverages your device's full potential with CPU, NPU, and GPU acceleration, supporting technologies like LiteRT and Gemini Nano for optimal performance.

主な機能

01Integrated LLMs (llama.cpp, GGUF), Vision AI, and Image Generation (stable-diffusion.cpp)
02CPU/NPU/GPU Hardware Acceleration (LiteRT, Gemini Nano, Snapdragon, Google Tensor)
03Private On-Device AI Suite with Hard Offline Mode
04Biometric Lock, Encrypted Chat History, and Tap-Jacking Protection
05Voice Mode (Speech-to-Speech AI Chat) and Document Analysis (RAG)
06483 GitHub stars

ユースケース

01Conducting private AI conversations and tasks entirely on an Android device.
02Generating images, analyzing documents, and leveraging vision AI locally without cloud dependency.
03Securely managing AI interactions with biometric authentication and encrypted history on any Android device, including custom ROMs.