Search for Nano Banana Pro pricing and nearly every result opens with a number. None of them comes from Google.
Nano Banana Pro pricing: the short answer
No official page this review could open states a price per image. The model page, which names the product Gemini 3 Pro Image, covers resolution, text rendering and limitations without mentioning cost. The announcement post discusses quotas rather than rates, and the developer surfaces are described as a paid preview.
The pages that would settle the question did not load here: the Gemini API pricing page, the image generation documentation on ai.google.dev, and the Vertex AI pricing page all timed out. So this article prints no unit price. Silence from an unreachable page is not a number.
What Google publishes about money, and what it keeps back
One sentence carries the entire free-tier policy.
“Our free-tier users will receive limited free quotas, after which they will revert to the original Nano Banana model. Google AI Plus, Pro and Ultra subscribers receive higher quotas.”
The free-tier sentence, read carefully
It claims three things: limited free quotas exist, running out drops you to the earlier Nano Banana model rather than locking you out, and subscribers of the three Google AI plans get more. There is no count in it, and the word price never appears. A quota limits how much you generate; a price says what a unit costs.
The subscription route: the only official prices we could read
In its May 2026 update, Google said it was launching a $100 monthly Google AI Ultra plan and cutting the monthly price of its top-tier Google AI Ultra plan from $250 to $200.
Those are prices for a plan, not for an image. Subscribing moves you into a higher quota band, and it never converts into a fixed number of renders. What the free route includes is in the Nano Banana Pro free guide.
The developer route: paid preview, with no figure attached
The developer announcement is precise about the shape of the deal and silent about the amount.
“This state-of-the-art image generation and editing model is starting to roll out in paid preview to build a new wave of intelligent, multimodal applications with the Gemini API in Google AI Studio and Vertex AI for enterprises.”
The only cost sentence Google attaches is relative: use Gemini 2.5 Flash Image for faster, lower cost image generation, or Gemini 3 Pro Image for higher quality generation, with higher cost and latency. That is a direction, not a rate. The Nano Banana Pro API guide covers what the developer pages decline to disclose.
Worked example: one poster prompt priced three ways
Here is the prompt, followed through all three routes.
“A travel poster for an overnight train from Zürich to Vienna. One carriage on a viaduct under a dark blue sky, warm lit windows, mountains behind. Headline text reading ‘NIGHT TRAIN’ across the lower third in a bold condensed sans-serif, and the route ‘Zürich – Vienna’ beneath it in small letters. Keep both text blocks crisp and legible. Aspect ratio 3:4, 2K resolution.”
It produces one 2K image with the headline spelled as written, the capability the model page promises for posters and diagrams. That page also lists the caveats: small faces, accurate spelling, fine details.
| Route | What the vendor publishes | What one 2K poster costs |
|---|---|---|
| Gemini app, free tier | A limited free quota, with no count | Nothing today; the ceiling is unpublished |
| Gemini API in Google AI Studio | A paid preview, at a higher cost than the earlier image model | No per-image rate is published |
| A reseller credit plan | That site’s own credit price, in that site’s own units | About $0.15 to $0.16 at the two rates we read |
Doing the arithmetic with a site’s own numbers
Two reseller pages opened for this review publish credit systems that look incomparable until you divide. The first shows a 12-credit render at 2K and states on its own plan card that 100 credits is roughly $1.25: about fifteen cents a poster. The second sells 15,000 credits a year for about 750 images at $118.80 a year, or twenty credits an image. Both are that site’s own quote.
What a paid tier changes about the output
Paying changes access and limits, not the model. Output at 1K, 2K or 4K, consistency for up to five characters, fidelity for up to fourteen objects: those hold on every surface.
The watermark is part of the price
This part of the bill arrives inside the file rather than on an invoice.
“In addition to SynthID, we will maintain a visible watermark (the Gemini sparkle) on images generated by free and Google AI Pro tier users, to make images even more easy to detect as Google AI-generated.”
Google removes that mark for Google AI Ultra subscribers and inside the Google AI Studio developer tool, and never removes SynthID. On client work it travels with the file.
How to read a credit price before you pay
- Check that the page names Gemini 3 Pro Image. A site that cannot name the model cannot tell you what you are buying.
- Divide the plan price by its credits, so every site lands in one unit.
- Check whether credits are monthly or annual. A yearly figure sold as a monthly price is this category’s oldest trick.
- Find the credit cost of the resolution you need. 4K usually costs more credits than 1K.
- Check whether the free allowance renews or arrives once.
If a page claims to quote Google’s official price, ask for the official page. There was none to link here.
Frequently asked questions
How much does Nano Banana Pro cost? No per-image price appears on the pages this review could open. The free tier runs on limited quotas, and the paid route is a plan, not a meter.
Is there an official API price? Not one verifiable here. The developer announcement calls the API a paid preview and gives only a relative cost.
Why does every site quote a different figure? They quote their own credit systems. One page presents $0.134 for a 1K or 2K image and $0.24 for 4K as official; another quotes a price in a different currency.
Is the free tier really free? Yes, with two conditions. The quota is limited and its size is unpublished, and free output carries the visible Gemini sparkle watermark as well as SynthID.
Which route should I price first? Start where you already work. The getting-started guide walks through a first image, and the topic hub collects the model and prompting pages.
The number nobody published is the number most pages claim to know.