How To Run the Latest Google AI Models on Your Phone for Free

Learn how to download and locally run Google’s Gemma AI model on your phone for free. You will set up a local AI chat that runs on your phone’s GPU, with no account, subscription, API key, or internet required after setup. The source guide frames this as a three-step setup.

  • Beginner
  • 3 steps
  • By The Rundown University

Rundown Pro

Every course and recording, the certification program and member perks, for $29 a month. Guides stay free.

Go Pro for $29

Step by step 3 steps

  1. Step 1 of 3

    Download the app

    Open the App Store or Google Play Store and search for Google AI Edge Gallery. Download it. It is free and does not require a sign-in.

    125-app-home-screen

  2. Step 2 of 3

    Download a model

    Open the app and tap AI Chat. You will see a few model options. Choose the recommended option. In the source example, the recommended model was Gemma 4 E2B, about 2.5 GB. Tap to download it and give it a minute.

    If you want something more capable, you can choose a bigger model. Tap Learn more next to any option to see what it does differently. There are also smaller models around half a gigabyte if you want a lightweight backup that takes up less space.

    Once the download finishes, you do not need internet again. The model lives on your phone.

    125-model-selection-download

  3. Step 3 of 3

    Enable Thinking and start chatting

    Before you send your first message, tap the settings icon in the top right of the toolbar and enable Thinking. This lets you see the model’s step-by-step reasoning before it gives you the final answer, while running entirely on your phone.

    You will also see other developer settings, such as max token limits and temperature. Since the model is running locally, you can experiment without usage cost.

    Start a conversation. You can try a travel-planning prompt like the example below.

    Plan a two-week road trip down Route 66 with landmarks, lodging, and food for each day.
    125-chat-response-thinking-mode

    One important caveat: this app is built for developers to test Google’s latest models before putting them in their apps. It does not save your chat history. If you close the app, the conversation is gone. You can tap the + button to see messages you have sent, but there is no way to pick up where you left off. Treat it as a scratch pad for one-off questions, not a notebook.

That's the whole workflow

Mark it complete to keep your progress. The background follows.

Go further with Google

These Rundown Pro recordings cover the same tool or topic as this guide.

$29/mo, cancel anytime

Background and context

Who This Is Useful For

  • Travelers who want a working AI on a plane, in another country, or anywhere without reliable wifi or cellular service
  • Privacy-focused users who do not want their prompts sent to a cloud server
  • Anyone curious about local AI who wants a simple first step into running models on their own hardware

What You Will Build

You will build a local AI chat running Google’s Gemma model directly on your phone with Thinking mode enabled. It runs without cloud access, an API key, or ongoing cost.

  • A downloaded local Gemma model in Google AI Edge Gallery
  • An offline-capable AI chat on your phone
  • Thinking mode enabled for chat responses

What You Need

  • An iPhone running iOS 17 or later, or an Android phone running Android 12 or later
  • An internet connection for the initial download only

Going Further

  • After setting up chat, you can explore more local AI uses. The app also includes audio transcription, image recognition, games, and Agent Skills such as a password generator, QR code creator, Wikipedia lookup, and interactive map. You can also search “Local AI” in Rundown University for more advanced guides on running your own AI models.