Google has quietly launched a new app called The AI ​​Edge Gallery, and its purpose is to change how artificial intelligence works on the smartphone. The app allows Android users to run powerful AI models directly on their devices, completely offline. In other words, you can now generate images, write code, or get smart answers without the need to connect to the Internet.

The main advantage of this step has increased privacy and rapid performance. Since everything runs on the device itself, there is no need to transmit data to the cloud server, reducing the risk of safety violations. This also means that there is no wait for the server to respond – AI answers immediately.

At the core of this experience is a language model called Jemma 3 1B. Weighing only 529MB, this compact model can process up to 2,585 tokens per second, which can lead to rapid text generation and uninterrupted interaction. Regardless of its small size, Gemma is powerful enough to support everything from custom material manufacturing to document analysis and smart answers in messaging apps.

One of the most reliable sources of the app Open AI model, also attracts the embrace facial, and is built on Google’s AI Edge Platform. This means that it benefits from techniques such as Tensorflow Lite and Medapipe, which helps customize performance in a wide range of equipment, even with minor hardware. He said, Google has said that the performance can be different based on the device. Older or middle-rang phones can struggle with large models, so the selection for lighter models is appropriate in such cases.

According to Google, users will find the interface refreshed. Features such as AI chat and Ask image provide a comfortable access to AI tools, while the Prompt Lab allows people to experiment with small, single-turn prompts. The lab also includes predetermined templates and settings to respond to the model.

Although the app is still that Google says a “experimental alpha release”, it is a completely open source under Apache 2.0 license. This means that developers and companies are equally free to use it, modify it, or even integrate it into commercial products. A iOS version is also reportedly on the horizon.

This development comes at a time amidst the investigation of Google’s broad AI ambitions. Last week, the US Department of Justice opened a civil antitrust inquiry into the company’s licensing deal with AI Startup Character .AI. Critics have expressed concern that the agreement may have been designed to dodge a federal merger review.

Despite this, the release of AI Edge Gallery is not only a leader in making Google offline AI not only possible but in practical. And with the ability to run the right powerful model from your pocket, it is a step towards AI more personal, private and always preparing-no Wi-Fi is necessary.