Introduction to the Google AI Edge Gallery APP
Google AI Edge Gallery is an experimental open-source app launched by Google that supports running generative AI models (such as Gemma 3n) entirely offline on Android/iOS devices. Users can locally deploy Hugging Face models to enable features such as AI chat, image question answering, audio transcription, and code generation, processing data without an internet connection and ensuring privacy and security. The app provides real-time performance monitoring (such as TTFT and decoding speed) and supports on-device Function Calling, turning a phone into a privacy-first AI assistant. It is suitable for developers exploring the potential of on-device AI and for privacy-conscious users in everyday use.
Main features of the Google AI Edge Gallery APP
- On-device offline inference: Supports running generative AI models (such as Gemma 3n) entirely offline on mobile devices, enabling AI chat, question answering, and code generation without a network connection.
- Multimodal interaction: Supports multiple task modes such as text conversation, image understanding and question answering (asking questions by inputting images), audio file transcription, and code generation.
- Model discovery and management: Connects directly to the Hugging Face platform to browse, filter, and download open-source models suitable for on-device operation to the local device.
- Privacy-first design: All data processing is completed locally on the device, and sensitive content does not need to be uploaded to the cloud, protecting user privacy and data security to the greatest extent.
- Real-time performance monitoring: Built-in metrics panel displays key performance data in real time, such as time to first token (TTFT), decoding speed, and memory usage.
- On-device Function Calling: Supports the model calling local tools or functions on the device to enable smarter automated task processing.
- Cross-platform support: Supports both Android and iOS systems, providing a native mobile experience.
