Google Releases AI Edge Gallery: Running Artificial Intelligence on Phones
In May 2025, Google quietly introduced a new application called AI Edge Gallery, which allows users to download and run artificial intelligence models directly on their smartphones. This experimental alpha application eliminates the need for an internet connection and cloud services when using generative artificial intelligence.
The application is currently available for Android devices and can be downloaded directly from GitHub under the Apache 2.0 license, which permits broad use, including in commercial applications. An iOS version is planned for the near future. Google describes AI Edge Gallery as an experimental alpha release and is actively encouraging developers to provide feedback to help further improve the application.
A key feature of AI Edge Gallery is its ability to search for, download, and run AI models directly from the well-known Hugging Face platform. This allows users to perform a wide range of artificial intelligence tasks—from generating images and answering questions to writing and editing code—all without requiring an internet connection. This offline functionality not only addresses privacy concerns by ensuring that user data is not sent to remote servers, but also ensures that AI features remain available even without Wi-Fi or a mobile connection.
The application includes a feature called "Prompt Lab," which offers templates for one-off tasks such as summarizing or rewriting text. Users can fine-tune model behavior using configurable settings, giving them greater control over the results of AI operations. The speed and efficiency of running models depend on two key factors—the performance of the device's hardware and the size of the AI model. More powerful phones naturally deliver faster results, while smaller models run faster than their larger counterparts.
