OpenModels

Nano Banana Pro, online

Google's image model, and the one to reach for when the picture has to contain readable text — a poster, a label, a sign — or when a face has to survive being restaged. It takes up to six reference images, so an identity you have already generated can carry into a new scene.

Which plan: Included from Starter.

What it is good at

Words inside the picture

Signage, packaging and UI copy come back legible instead of as letter-shaped noise.

Six references, one identity

Feed it the same character from several angles and it keeps the face across a new scene.

Long prompts hold

Detailed art direction survives the whole way through rather than collapsing to the first clause.

How to prompt this model

Use clear scene descriptions and precise editing constraints.

Example prompt

Create a square bakery poster with a single croissant on cream paper. Place the exact heading "Fresh at dawn" above it in dark blue, with generous space around the lettering.

Read the prompting tutorial. Select this model before using Enhance prompt to apply its guidance to your draft.

Sources: Google image generation guide.

The specifics

SettingDetails
Made byGoogle
RunsText to image, and editing from references
Reference imagesUp to 6
FramingSquare, portrait, widescreen and four more
Opens atStarter

A prompt to start from

A weathered enamel shop sign reading OPEN MODELS in hand-painted serif letters, hanging above a rain-slick cobbled street at dusk, warm window light behind it, shallow depth of field

Open Nano Banana Pro in the studio — Choose the model and review your prompt and quote before generating.

Questions people ask

Can I use Nano Banana Pro for free?

Nano Banana Pro opens at the Starter plan. GPT Image 2 is open on every plan, free accounts included — the first four runs of the day are ⚡1 from the daily wallet, then ✦4 in Premium model credits — and it runs from the same prompt bar.

Can I run Nano Banana Pro and other models on one prompt?

Yes. Pick several models in the prompt bar and one press sends the prompt to all of them; the results arrive together in a single album so you can compare them side by side.

Does it edit images or only generate them?

Both. Attach up to six reference images and the prompt becomes an edit instruction against them instead of a description from scratch.

Compare models