Last week, Google quietly released an app, allowing users to run a series of AI models openly available from the AI Dev platform Hugging Face on its phone.
called Google AI Edge GalleryThe app is available for Android and will soon come in iOS. It allows users to find, download and run compatible models that generate pictures, answer questions, write and edit the code, and more. Models run offline, without the need for internet connection, tapping in a supported phone processor.
The AI models running in the cloud are often more powerful than their local counterparts, but they also have their own downsides. Some users may beware of sending individual or sensitive data to a remote data center, or want to provide models without the need to find Wi-Fi or cellular connections.
Google AI Edge Gallery, which Google is calling a “experimental alpha release”, can be downloaded by the following from Github These instructionsThe home screen shows shortcuts to AI functions and capabilities such as “Ask Image” and “AI Chat”. Tapping on a capacity leads to a list of models favorable for work, such as Google’s Gemma 3n.
Google AI Edge Gallery offers a “Prompt Lab”, which can be used to kick the “single-turn” functions operated by user models, such as summarizing and re-writing the text. The Prompt comes with several task templates and configuble settings to fix the behavior of the lab model.
Your mileage may vary in terms of performance, Google warns. Models will run fast before modern devices with more powerful hardware, but the size of the model also matters. The larger model will take longer to complete a task – says, answering a question about an image – compared to the small model.
Invite invited members of Google’s developer community to react to Google AI Edge Gallery Experience. The app is under the Apache 2.0 license, which means that it can be used in most contexts – commercial or otherwise – without a restriction.
