Run Google’s new Gemma 4 AI models locally on Android and iOS: Here’s how

Google has unveiled Gemma 4, a groundbreaking suite of open AI models designed to enhance mobile devices with advanced intelligence capabilities. What sets these models apart is their ability to function offline, allowing smartphones to harness powerful AI without the need for an internet connection. At the forefront of this initiative is the AI Edge Gallery app, which serves as the primary interface for users on both Android and iOS platforms. This innovative tool enables users to tap into the reasoning and adaptive capabilities of Gemma 4, marking a significant step in the development of AI applications that operate seamlessly on local hardware. Gemma 4 represents a leap forward in Google's Gemini research, featuring a range of lightweight models tailored for flexibility and performance. Unlike traditional closed systems, these models can be deployed across various devices, providing developers with the freedom to work in both local and cloud environments. The lineup includes four distinct variants: Effective 2B (E2B), Effective 4B (E4B), 26B Mixture of Experts (MoE), and 31B Dense, each catering to different processing needs. The Google AI Edge app is designed to streamline the development process, offering a comprehensive stack for deploying AI models on Android, iOS, and edge devices. Users can access advanced large language models (LLMs) and multimodal capabilities entirely offline, ensuring data privacy and security. Among the features of the AI Edge app are a conversational AI interface known as Thinking Mode, a dedicated workspace for text generation, summarization, and rewriting, as well as model management tools. Additionally, users can utilize functionality such as Ask Image for visual inquiries and Audio Scribe for real-time transcription and translation. To start using Gemma 4 on mobile, users should follow these steps: First, download the Google AI Edge app from the Google Play Store for Android or the App Store for iOS, with the recommendation of using an iPhone 15 Pro or newer for optimal performance. Next, open the app and choose from various modes, such as AI Chat, Ask Image, Audio Scribe, or Agent Skills. Then, navigate to the Models tab and select either the E2B model for speed or the E4B model for enhanced intelligence based on device capabilities. After downloading the chosen AI model, users can enjoy the full range of its features directly on their devices, free from major limitations.