Google has quietly introduced an application that allows users to run a variety of open-source AI models provided by Hugging Face on their mobile devices. Named the "Google AI Edge Gallery," this app is currently available on Android, with an iOS version forthcoming. Through this app, users can discover, download, and run compatible models for tasks such as image generation, question answering, and code editing. These models function offline, utilizing the device's processor without the need for an internet connection.
While cloud-based AI models are typically more powerful, they come with drawbacks. Users may be reluctant to send sensitive data to remote servers or may want to use models in environments without Wi-Fi or mobile networks. The Google AI Edge Gallery is labeled as an "experimental Alpha version" and can be downloaded via a GitHub link. The main screen provides shortcuts for AI tasks like "Image Q&A" and "AI Chat," displaying a list of applicable models for each task.
The app includes a "Prompt Lab" for initiating model-driven tasks such as text summarization and rewriting, offering templates and settings to fine-tune model outputs. Google notes that performance may vary based on device hardware, with modern, high-spec devices expected to run models faster.
