How to Use Gemini Nano Banana: A Complete Guide (2026)

Gemini Nano Banana is Google's image model for creating and editing pictures with plain-language prompts. Knowing how to use Gemini Nano Banana well comes down to three things: choosing the right version of the model, writing prompts that describe a scene instead of listing keywords, and using follow-up edits instead of starting over.

This guide covers each step, from your first image to character consistency, photo editing, and readable text. You will also see how to use Nano Banana and Nano Banana Pro on Chat Smith, an AI chat app that puts Google's image models next to GPT, Claude, Grok, and DeepSeek, so you can generate, compare, and refine images in one place.

Nano Banana started as a community nickname for Gemini 2.5 Flash Image, and Google adopted the name for its whole family of image models. Unlike older generators that create every picture from scratch, Nano Banana models understand the images you upload, so they can edit a photo, keep the same person across a series, or blend several references into one result.

There are now three versions worth knowing:

What sets Nano Banana apart is how well it follows instructions about existing images. The features you will use most:

• Character consistency: keep the same face, outfit, or product across multiple images.
• Targeted edits: change one part of a photo, such as the background or a shirt color, without disturbing the rest.
• Readable text: generate posters, signs, and labels with correctly spelled words, especially with Nano Banana Pro.
• Multi-image blending: combine the subject of one image with the style or setting of another.
• World knowledge: create infographics, diagrams, and recipe cards that reflect real facts, not just pretty shapes.

You can use Nano Banana through Google's own apps, through the Gemini API, or through a multi-model platform. Which one fits depends on how often you create images and whether you also need other AI models.

1. Open the Gemini app or web app and start a new chat.
2. Type a prompt that starts with an action such as “Create,” “Generate,” or “Edit,” or upload a photo you want to change.
3. Review the result, then refine it with follow-up messages in the same chat.
4. On a paid plan, use the redo option to regenerate an image with Nano Banana Pro when you need higher resolution or cleaner text.

The free tier is a good way to try the model, but limits are tight and change with demand. Heavy users usually end up on Google AI Pro or Ultra, which covers Google's models only.

1. Open Chat Smith on web, Android, or iOS.
2. Choose Nano Banana for quick images or Nano Banana Pro for detailed, high-resolution work.
3. Type your prompt or upload reference images, then refine with follow-ups.
4. Switch to another image model with the same prompt if you want a second option.

The advantage is choice without extra subscriptions. You can run a prompt through Nano Banana Pro, compare it with GPT Image 2.0, and keep whichever result fits better, all for less than a single Google AI Pro plan.

Nano Banana reads prompts the way a designer reads a brief. The more clearly you describe the image you want, the fewer rounds of edits you need.

A reliable structure is action, subject, details, style, and lighting. You don't need every element every time, but each one you add removes a guess the model would otherwise make.

Weak prompt:

Make a logo for a coffee shop.

Strong prompt:

Generate a minimalist vector logo for a coffee shop named “Bean Smith,” featuring a steam swirl shaped like a leaf, flat colors, white background, high contrast.

For brand work, generate several variations first and pick a direction before you refine. If you want a faster starting point for marks and icons, the AI logo generator is built for exactly that task.

Older image models rewarded comma-separated tags. Nano Banana is built on a language model, so full sentences work better: they tell it how the parts of the picture relate to each other, not just what is in it.

Keyword list:

cat, café, window, watercolor, warm, 4k

Scene description:

A ginger cat sleeping on the windowsill of a small corner café at golden hour, painted in soft watercolor with warm orange and cream tones. Rain streaks on the glass, a half-finished latte in the foreground.

The second prompt fixes the camera position, the mood, and the small details that make an image feel intentional. This works the same way whether you are using text to image for illustrations, product shots, or social content.

Mentioning the destination helps Nano Banana choose the right framing and leave room for text or cropping. Common formats:

For video covers specifically, the AI YouTube thumbnail maker applies these layout rules for you, so you only need to describe the subject and title.

Character consistency is Nano Banana's biggest strength. Instead of describing a person from scratch each time, you upload a reference and ask the model to keep it intact while changing everything else.

1. Upload a clear reference image of your character or product, ideally front-facing and well lit.
2. Refer to it directly as “this character” or “the product in this image.”
3. State what must stay the same, then describe the new scene, pose, or outfit.

Example prompt:

Create a new image of this character sitting in a café drinking tea. Keep the same face, hairstyle, and outfit. Soft morning light, shallow depth of field.

For a series, generate one image at a time in the same chat so the model keeps the context. If details start to drift, re-upload the original reference. The same approach powers image to image workflows for product catalogs, mascots, and comic panels.

Upload a photo and describe the change you want. Nano Banana edits only what you mention and leaves the rest of the image alone, which makes it far faster than manual retouching.

Make one change per message so you can undo it easily. For quick cleanups like cutting out a subject, a dedicated AI background remover is often the fastest route.

Nano Banana can spell, which makes it useful for signs, posters, and social graphics. Put the exact words in quotation marks and describe the font style and color.

Example prompt:

Create a neon sign on a brick wall that says “OPEN 24/7” in pink cursive letters, with a soft glow reflecting on the wet pavement below.

• Keep headlines short; a few words render far more reliably than full sentences.
• Say where the text goes: “top center,” “on the mug,” “on the storefront awning.”
• Switch to Nano Banana Pro for dense layouts, small print, or text in multiple languages.

If you are making event or promo material, the AI poster generator gives you a layout-ready starting point that you can then refine with your own prompts.

Upload two or more images and tell Nano Banana what to take from each. This works well for mockups, mood boards, and placing a product into a lifestyle scene without a photo shoot.

Example prompt:

Place the sneaker from the first image on the wooden table in the second image. Match the lighting and shadows of the second image, and keep the sneaker's logo and colors exactly the same.

Label each image by its role (“the product,” “the background,” “the style reference”) so the model does not mix up which elements to keep.

Great images usually start before the prompt: with research, a clear concept, and a well-written brief. On Chat Smith, every step happens in the same app:

1. Research: ask Grok 4.5 which visual trends are popular in your niche right now.
2. Write the brief: have Claude Sonnet 5 or GPT-5.6 Sol turn your idea into a detailed Nano Banana prompt.
3. Generate: run the prompt through Nano Banana Pro.
4. Compare: send the same prompt to GPT Image 2.0 and keep the stronger result.
5. Refine: polish the winner with follow-up edits in natural language.

Doing the same with official apps means paying for more than one subscription. As of September 2026, Google AI Pro costs $19.99 per month and ChatGPT Plus costs $20 per month, and each covers only its own company's models.

A good image is often only half of the job. Posters, social posts, and brand assets also need layout, typography, and a consistent look. Chat Smith Design Studio brings AI image generation and design tools into one workspace, so you can go from a prompt to a finished, ready-to-share design without exporting files between apps.

For single-purpose jobs, Chat Smith also offers focused tools such as the AI image generator for general visuals and the AI headshot generator for professional profile photos.

Learning how to use Gemini Nano Banana is mostly about writing clear briefs: describe the scene, name the format, upload references when consistency matters, and refine with small follow-up edits. Once those habits are in place, you can produce product shots, social graphics, and posters in minutes. With Chat Smith, you get Nano Banana and Nano Banana Pro alongside GPT Image 2.0 and leading chat models in one app, without stacking subscriptions.