Google AI Edge Gallery just launched an app that lets users run AI models from Hugging Face directly on their Android phones. iOS users will get access soon.
The app, called Google AI Edge Gallery, breaks away from traditional cloud-based AI. It allows users to find, download, and run models that can generate images, answer questions, and even write code—all offline.
Big name AI models often need cloud support for power, but many users worry about data privacy and connectivity issues. This app mitigates those concerns by running models on-device. 
The app is currently labeled as an “experimental Alpha release,” available on GitHub. Users can access a home screen rich with AI tasks and capabilities like “Ask Image” or “AI Chat.” Tapping a capability reveals model options tailored for that specific task, such as Google’s own Gemma 3n.
A “Prompt Lab” feature lets users initiate tasks like summarizing or rewriting text. It includes task templates and settings to customize model behavior.
Performance can vary. More powerful devices will execute models faster, but larger models might slow down task completion—like answering questions about images.
Google is reaching out to developers for feedback on the experience. The app is under an Apache 2.0 license, allowing broad usage without restrictions.
"Your mileage may vary in terms of performance," Google warns.
"Modern devices with more powerful hardware will predictably run models faster."
Google encourages experimentation and engagement, marking another step in making AI access more personal and immediate.
