AI on your smartphone is a wild idea but what happens when there’s no internet connectivity? AI models like Gemini 2.5 Flash and OpenAI ChatGPT rely on internet connectivity to do their AI things. The core reason behind this is the limited AI computing power available on smartphones and laptops – AI models need dedicated GPU power to run complex models and execute calculations. Google, however, has cracked something on this front – to make AI run on your phone without data connectivity.
Away from the usual AI fanfare, Google has quietly released an app that runs AI models on your device. Called the Google AI Edge Gallery, this app runs AI models locally and can execute basic generative AI tasks like chatting and image generation. It can also answer questions, generate code, and even edit pre-written code.
One of the most crucial advantages to running AI models locally is the preservation of data privacy, something that online models aren’t capable of.
How does Google AI Edge Gallery work?
Think of this app as a platform to run third-party AI models available on AI dev platform Hugging Face. The app’s homescreen offers three separate sections – Ask Image, Prompt Lab and AI Chat.
When users tap on Ask Image and AI Chat, the app will show them all the available AI models suited to their needs. The Prompt Lab section allows one to “summarize, rewrite, generate code, or use freeform prompts to explore single-turn LLM use cases.” With Ask Image, you can “upload an image and ask questions about it. Get descriptions, solve problems, or identify objects.”
Similarly, the AI Chat section lets you have a turn-by-turn conversation with the AI model.
How to download Google AI Edge Gallery?
At the moment, the Google AI Edge Gallery is part of Google’s experimental alpha release and hence, may not be as stable as online AI models. As a result, the app is available only via GitHub and can be downloaded for Android phones only at the moment. A version for iOS is also slated to be released soon.