Google AI Edge Gallery lets users run AI models offline on Android phones, enabling tasks like image generation and code writing without internet.
Google has launched a new app called AI Edge Gallery, designed to let users run lightweight AI models completely offline on their smartphones. Initially available for Android via GitHub, the app allows users to perform tasks like image generation, code writing, and smart responses — all without needing an internet connection.
The core benefit of this approach is enhanced privacy and faster performance. Since processing happens locally on the device, there’s no need to send data to cloud servers, reducing privacy risks and delivering quicker results.
Powering this offline AI experience is Gemma 3 1B, a compact language model developed by Google. Despite its size of just 529MB, it can process up to 2,585 tokens per second. This allows it to handle tasks like content generation, summarisation, document analysis, and smart replies — all on-device.
The AI Edge Gallery app integrates third-party models sourced from Hugging Face, a widely trusted platform for open-source AI models. Built on Google’s AI Edge platform, it leverages tools like TensorFlow Lite and MediaPipe to optimise AI performance across a range of devices, including those with limited hardware resources. However, performance may vary — older or mid-range smartphones may perform better with lighter models.
Google Gemini Now Automatically Summarizes Gmail Emails
The app offers a user-friendly interface featuring sections such as AI Chat, Ask Image, and Prompt Lab. AI Chat supports conversational interactions with AI models, while Ask Image allows users to upload photos and get descriptions, object identification, or problem-solving suggestions. Prompt Lab is aimed at quick, single-turn tasks like summarising, rewriting, or generating code, and also includes prebuilt templates to help users get started.
Although still in what Google calls an “experimental Alpha release”, the app is open-source under the Apache 2.0 licence, which means developers and businesses can use, modify, or integrate it into their own products. An iOS version of the app is also expected soon.
This launch marks a significant step for Google in bringing offline AI capabilities to personal devices. Unlike services like ChatGPT or Gemini Flash, which require constant cloud connectivity, AI Edge Gallery runs everything natively on your device. That not only boosts privacy but also eliminates dependency on internet access — making AI more personal and always available.
