Nano Banana
#Figurine
The 1/7 figurine trend, done in one prompt. Character on an acrylic base, the sculpt on the monitor, the box art beside it. Swap in your own character and it keeps the look.

Type a prompt or drop in a photo. Nano Banana 2.1 and 40+ image and video models are one click away, and every result is saved to your library.
Core features
Start from words, a photo, or a pair of frames. The form only shows what the chosen model supports.
Describe the scene in plain language — subject, light, composition, even the words on a sign — and get up to four takes at once.

Add reference photos and say what should change: swap the background, restyle an outfit, or merge several shots while keeping the subject recognizable.

Switch to video for Veo 3.1, Kling 3.0, Seedance 2.0, Wan and more. Animate from text, from a first and last frame, or from reference images.

Finished images and videos are copied to our own storage and listed under My works, so links keep working long after the job is done.

Why it works
The parts you notice after the tenth generation, not just the first.

Google, OpenAI, ByteDance, Kling, Alibaba, xAI and more — searchable by vendor, with sensible defaults for each.

Feed the same person, product or character in several references and Nano Banana keeps their look across edits.

Posters, labels and covers with words that read correctly.

The exact credits for this model and these settings appear before you press Generate.

Each image or clip is charged on its own. If one fails, its credits come back automatically.

Write in the language you think in. The interface is available in English, Chinese, Japanese and Korean.

Reference images and results are used only to fulfil your request and stay in our own storage. We never use them to train models of our own.
Toolkit
Every tool opens the same generator, preset for the job.
Text to imageFrom a sentence to a finished picture.Try it
Photo editingChange a photo with words and references.Try it
Batch variationsUp to four images per prompt to compare.Try it
Text to videoShort clips with optional sound.Try it
First and last frameSet where a shot starts and ends.Try it
Reference to videoKeep a character or product in motion.Try itTrending
The styles everyone is sharing right now. Pick one and make your own in a click.
Nano Banana
#Figurine
The 1/7 figurine trend, done in one prompt. Character on an acrylic base, the sculpt on the monitor, the box art beside it. Swap in your own character and it keeps the look.

Nano Banana
#Sketch to plush
Sketch → 3D plush. A quick doodle on notebook paper becomes a soft, stitched toy you could almost pick up.

Nano Banana
#Product shot
No studio, no photographer. One product photo in, a luxury ad shot out — wet volcanic rock, frozen splashes and golden rim light.

Nano Banana
#Photo restoration
Old family photos, brought back. Scratches and creases removed, faces sharpened and colors added naturally — without changing who they are.

Nano Banana
#Miniature diorama
A whole street corner on a tiny base. Isometric miniatures are the easiest way to make a place look magical.

Nano Banana
#Poster with text
Text that actually reads. Title, credits and layout in one go — handy for covers, posters and thumbnails.

Pricing
Subscribe for credits every billing period, or buy a credit pack once. One kind of credit works with every model.
Cancel anytime. Partial periods are not refunded.
Starter
-50%Billed yearly · $107.88/year $215.88
For trying things out
Pro
-50%Billed yearly · $215.88/year $431.88
For regular creators
Premium
-50%Billed yearly · $431.88/year $863.88
For heavy video and 4K work
FAQ
Short answers before you start.
Nano Banana is the name of Google's image generation and editing models. This site offers Nano Banana 2.1, Nano Banana 2 and Nano Banana Pro alongside image and video models from other vendors. We are an independent service and not affiliated with Google.
Each model and combination of settings has a fixed credit cost, shown in the generator before you submit. Images are counted per image; videos per clip, and longer or higher-resolution clips cost more.
Every image or clip is settled on its own. Outputs that succeed are charged; the ones that fail or time out are refunded automatically.
Results are copied to the site's own storage and listed under My works once you are signed in, where you can view them again later.
JPEG, PNG or WebP images up to 10 MB each. How many references you can add depends on the model and is shown in the generator.
Yes. Most models understand both Chinese and English. When you need very precise control, an English prompt is often a little more predictable.
Nano Banana, GPT Image, Seedream, Veo, Seedance and more in one place. Type an idea and see it in seconds.