Nano Banana in Chrome & Gemini: How It Works

Google’s Nano Banana image model, powered by the Gemini family, is quietly becoming part of everyday tools like Chrome and the Gemini app. This guide explains what Nano Banana is, how it’s being integrated, and what you can do to see it in action or control its presence on your devices.

What You’ll Need

A compatible device: a recent Android phone (Pixel 9 series or newer recommended) or a Chromebook Plus model.
Google Chrome browser updated to a recent version (check via chrome://settings/help).
The Google Gemini app installed and signed in with a personal Google Account.
An internet connection for the initial model download and for cloud-based features.

Step-by-Step

Understand what Nano Banana is. Nano Banana is the informal name for Gemini’s on-device image generation and editing model. It runs locally on your hardware rather than in the cloud, which means faster responses, offline capability, and better privacy for your images. It uses the same underlying technology as the larger Gemini image models but is distilled to fit on a phone or laptop.
Check if your device supports it. Nano Banana is currently rolling out to Pixel 9 and Pixel 10 series phones, and to Chromebook Plus laptops with compatible hardware (typically those with a Google Tensor or a recent Intel/AMD chip with sufficient RAM). If you have an older device, you may not see the features at all.
Enable on-device AI in Chrome. Open Chrome on your Android phone or Chromebook. Go to chrome://flags in the address bar. Search for “on-device AI” or “Gemini Nano”. Look for flags like “Enable Gemini Nano on-device AI” or “On-Device AI Model” and set them to Enabled. Restart Chrome when prompted. Note: flags can change or disappear between versions, so if you don’t see them, your device may not be in the rollout group.
Use Nano Banana in Chrome’s built-in tools. With the flag enabled, Chrome can use Nano Banana for features like “Compose” (a writing assistant) and, more recently, for image editing in the context menu. To test, right-click (or long-press) on any image in a web page. If the feature is active, you’ll see an option like “Edit with Gemini” or “Generate image”. Selecting it will open a small panel where you can type a prompt like “make the sky more dramatic” or “remove the background”. The model processes it on-device and returns an edited image.
Access Nano Banana in the Gemini app. Open the Gemini app on your phone. Tap the “Gemini” icon or the “+ Create” button. Look for an option called “Image” or “Create image”. If you see a badge that says “On-device” or “Nano”, that means the local model is being used. You can then type a prompt such as “a cat riding a skateboard” and the image will be generated without sending your prompt to a server.
Manage the model download. Nano Banana is downloaded as a separate package, usually a few hundred megabytes. On Android, you can manage this via Settings → Google → Device & sharing → On-device AI. Here you can see the download status, update the model, or delete it to free space. On Chromebook, look in Settings → About Chrome OS → Additional details or check the same Google settings page if available.
Test offline capability. Once the model is downloaded, you can try generating an image with your device in airplane mode. If it works, you’ve confirmed Nano Banana is running locally. If it fails, the feature may still be using a cloud fallback.
Keep an eye on updates. Google is rolling out Nano Banana gradually. To get the latest features, keep Chrome and the Gemini app updated via the Play Store or your device’s update settings. Also check the chrome://flags page periodically for new AI-related flags.

Common Mistakes

Expecting it on all devices. Nano Banana is not available on every phone or laptop. It requires specific hardware (Tensor chip or recent CPU with enough RAM). If you don’t see the options, your device simply isn’t supported yet.
Confusing cloud Gemini with Nano. The Gemini app often defaults to cloud-based processing. Look for explicit “on-device” or “Nano” labels. If you’re not sure, try airplane mode – if it fails, you were using the cloud.
Editing flags without knowing what they do. Changing flags can break Chrome or cause instability. Only change the specific flags mentioned here, and reset them if you encounter issues.
Assuming all image editing is Nano. Chrome’s right-click image edit might still use a cloud model for complex tasks. Nano is used for simple, on-device edits; Google decides which tasks go local vs cloud.

Tips

Use clear, simple prompts for best results. Nano Banana is a smaller model, so it works better with short, direct descriptions like “a red car” rather than complex scenes.
Keep your device charged and cool. On-device AI can drain battery quickly, especially during image generation.
If you’re a developer, keep an eye on Google’s AI Edge documentation for APIs to integrate Nano Banana into your own apps.
Regularly clear the model cache if you’re running low on storage. The on-device AI settings page usually has a “delete model” option that will re-download when needed.

Troubleshooting

Feature not showing up. First, verify your device is on the supported list. Then check Chrome flags are enabled and restart the browser. If it still doesn’t appear, your Google Account might not be in the rollout group – wait a few weeks or try a different account.
Model download fails. Ensure you have a stable internet connection and at least 500MB of free storage. Go to the on-device AI settings and tap “Retry” or “Update”. If it keeps failing, restart your device and try again.
Image generation is very slow. This is normal for on-device models, especially on older hardware. Close other apps to free up memory. If it’s unbearably slow, the feature might be using a cloud model anyway – check the network activity in your device’s data usage settings.
Chrome crashes after enabling flags. Go back to chrome://flags and set the AI flags back to Default. Restart Chrome. If crashes persist, clear your browser cache or reinstall Chrome.
Edited images look low quality. Nano Banana is a compact model, so it may not match the quality of the full Gemini image model. Try using more specific prompts or edit at higher resolutions. If you need professional quality, use the cloud-based Gemini in the app.

Frequently Asked Questions

Is Nano Banana the same as Gemini?

No. Nano Banana is a distilled, on-device version of Google’s Gemini image model. It runs locally for speed and privacy, while the full Gemini model runs in the cloud and offers more advanced capabilities.

Will Nano Banana work on my iPhone?

No. Nano Banana is designed for Android and ChromeOS devices. iOS users get cloud-based image generation through the Gemini app, but not the on-device version.

Does Nano Banana work offline?

Yes, once the model is downloaded, basic image generation and editing work offline. However, some features like sharing or saving to Google Photos may require an internet connection.

Is my data private when using Nano Banana?

Yes, that’s the main benefit. Since the model runs on your device, your prompts and images are not sent to Google’s servers for processing. However, if you choose to use cloud-based features, your data will be handled according to Google’s privacy policy.

How do I know if I’m using Nano Banana or the cloud model?

Look for a small badge or label that says “On-device” or “Nano” in the Gemini app or Chrome’s edit panel. Alternatively, turn on airplane mode and try to generate an image – if it works, you’re using Nano.

Can I disable Nano Banana?

Yes. On Android, go to Settings → Google → Device & sharing → On-device AI and turn off the model or delete it. In Chrome, you can disable the flags you enabled. This will stop the on-device features, and Chrome/Gemini will fall back to cloud-based AI.

General steps for current mainstream versions; menus can differ by device and version. See the disclaimer or report an error.

More in this bed

The whole Tech How-To bed →