How to Use Gemini's Image Generator (and Edit Photos in Gemini)

To use Gemini's image generator, open the Images page in the Gemini app or at gemini.google.com and type what you want to see. By default, Gemini draws those images with Nano Banana 2 (Gemini 3.1 Flash Image), the model Google launched on February 26, 2026. The sections below cover the computer and phone steps, photo editing, the model setting to choose, and the free tier.

4 ways to use the Gemini image generator

Generating images requires a signed-in Google Account and an age of 13 or more (or your country's minimum). As of September 2026, there are four routes:

How to use Gemini to create images on a computer

The web version has the same Images page as the app, with a full-size download.

1. Go to gemini.google.com and sign in.
2. Click Open Sidebar at the top left, then Images.
3. Choose a template, or type a prompt that starts with "Create an image of", "Generate" or "Draw".
4. Ask for changes in plain words in the same chat, one change at a time.
5. To save the picture, hover over it and click Download full size.
6. To keep working in a document, click More under the image, then Export to Docs.

Image generation only works in the countries and languages where the Gemini app is offered, and school or work accounts can carry extra restrictions.

How to use Gemini to generate pictures on your phone

The Gemini app works the same way on Android and on iPhone or iPad.

1. Open the Gemini app.
2. Tap Menu at the top left, then Images.
3. Start from a template, or write your own prompt.
4. On iPhone or iPad, tap the result to see it larger, then tap Back to return to the chat.
5. To keep it, touch and hold the image and tap Save. Share creates a public link, so skip it for anything private.

Gemini Live works too: start it and ask for an image out loud, or share your camera in Live and ask for an image based on what it sees. On Android, if Gemini is your main assistant, you can also say "Hey Google" to start.

How to use Gemini to edit photos

Gemini edits a photo from a written instruction, and only users 18 or older can edit.

1. Open the menu or sidebar and select Library.
2. Select an image you made in Gemini, or upload one from your device.
3. Under the image, select Chat.
4. Type the change in the box at the bottom, then select Submit.

Say what changes and what stays, for example "Change the sofa to deep green velvet and keep the rug, the lamp and the walls as they are." You can also upload several images and ask for one new picture from them: "Put the dog from the first photo on the beach from the second, sitting on the sand at sunset." For more instructions to copy, see the photo editing prompts.

In the US, adults with a personal account can connect Google Photos and edit pictures from their library. On the web, go to Settings and help, then Personal Intelligence, then Connected Apps, and turn on Google Photos. Gemini uses the names on your Google Photos face groups to find people and pets, so name a face group before you ask for a picture of that person.

Which Gemini model setting makes your image

The model you pick in Gemini's model menu decides which image model runs, so check it before you type a long prompt.

Some Google pages label the menu settings Fast, Thinking and Pro instead; in Google's help center, the Lite image model runs only when Flash-Lite is selected. To redo an image, create it first in Nano Banana 2 or its Lite version, then click or tap More at the bottom right and choose Redo with Pro. If you've used up the day's Nano Banana 2 images, Redo with Pro stops working too. For the background on each model, read what Nano Banana is.

Prompts for the Gemini image generator

To use Gemini to generate images close to the one in your head, give it a subject, an action, a setting and a style, then add the shape. Google's tips are to open with an action word like draw, generate or create, to name a style such as photorealistic, charcoal drawing or watercolor, and to state the aspect ratio in the prompt. These three work in Gemini as written, and in other image tools.

Create an infographic poster that explains how to care for a sourdough starter in four numbered steps, labeled "Mix", "Feed", "Wait" and "Bake". Flat illustration style, warm cream and brown colors, a tall 3:4 layout, clear sans-serif labels, no other text.

Generate a photorealistic image of a sleepy golden retriever puppy curled on a knitted blanket beside a fireplace. Low camera angle, warm firelight from the right side, a softly blurred background, wide 16:9 frame.

Draw a quick, loose watercolor of a small blue fishing boat tied to a wooden dock at sunrise, soft pink sky, visible brushstrokes, lots of white paper showing, square 1:1 format.

Google says Nano Banana 2 pulls in current information and pictures from Google Search results, which helps with infographics, diagrams and real-world subjects, so name the landmark, building or plant species you mean. For more to copy, browse the Nano Banana prompt library. The steps and prompt habits are similar in ChatGPT, covered in how to make images with ChatGPT.

Free limits and Google AI plans

Nano Banana 2 image generation is included for everyone, with or without a paid plan. The limits are compute-based: longer chats, heavier models and media generation use more of your allowance. It refills every 5 hours until you hit a weekly limit, and your status is under Settings, then Usage Limits.

What each Google AI tier costs and adds, for US accounts in September 2026:

Google also offers college students a free AI Pro plan for one year. Google warns that limits can change without notice and that free accounts may be cut back first when capacity is tight.

Does Gemini put a watermark on images?

Yes, usually two. Google says Gemini images carry SynthID, an invisible mark that identifies them as AI-generated, plus a visible watermark. When it launched its Pro image model in November 2025, Google said images from Ultra subscribers would skip the visible one. To check a picture someone sent you, upload it to Gemini and ask if Google AI made it. The Nano Banana explainer covers Google's Content Credentials too.

Troubleshooting Gemini images

Run the same prompt in Image Generator

The three prompts in this guide paste straight into Image Generator in Toybox AI. Type the prompt in "Describe your image", choose one of the five shapes under "Size" (from Square 1:1 to Wide 16:9), and open "More options" to switch on "Enhanced quality" when the picture has words. To combine pictures the way Gemini does, tap "Add reference photos", add up to 3, and explain in the prompt which part of each photo to use. Faces copied from a reference photo can drift, so compare them with the original before you share. A picture is currently 50 credits (95 with "Enhanced quality"). A new account's 20 free credits won't cover one, so look at Toybox pricing first.

For a photo you already have, Smart Editor works from one upload and a plain-language request typed into "What should change?". You need Pro or a higher plan for it, and an HD edit is currently 85 credits. It can't merge two photos into one. Because every edit redraws the full picture, end your request with the words "keep everything else the same". Once a result appears, the "Editing" switch lets you pick "Latest result" to build on it or "Original photo" to start again from your upload.