petalumi.
Little illustrations for digital journals
petalumi.

Nano Banana Pro: Google's Studio-Quality Image Model, in Official Examples

Nano Banana Pro (Gemini 3 Pro Image) is Google's most capable image model. See Google's official examples and prompts for infographics, legible text, translation, 14-image blending and studio controls, plus specs and pricing.

3 min read

Nano Banana Pro is Google's name for Gemini 3 Pro Image, the image generation and editing model built on Gemini 3 Pro. Google introduced it on November 20, 2025 as its "new state-of-the art image generation and editing model," designed to use "Gemini's state-of-the-art reasoning and real-world knowledge to visualize information better than ever before." Every image on this page is an official example from Google, shown with the prompt Google used.

Ideas turned into infographics

Built on Gemini 3's reasoning, Nano Banana Pro can create "context-rich infographics and diagrams based on the content you provide or facts from the real world."

An infographic about the String of Turtles house plant, made from a photo of the plant
Prompt: “Create an infographic about this plant focusing on interesting information.” Image: Google
A step-by-step infographic for making elaichi chai
Prompt: “Create an infographic that shows how to make elaichi chai” Image: Google

Connected to Google Search

The model can use grounding with Google Search to work with real-time information. Google used it to pull in the live weather for a pop-art infographic.

A comic-book style weather infographic for Dallas, Texas, built from live weather data
Google: “We used Nano Banana Pro to pull in real-time weather via Search grounding to build a pop-art infographic.” Image: Google

The text is the design

Google calls Nano Banana Pro "the best model for creating images with correctly rendered and legible text directly in the image, whether you're looking for a short tagline, or a long paragraph."

Old houses on a Berlin street shaped like the letters of the word BERLIN
Prompt: “View of a cozy street in Berlin on a bright sunny day, stark shadows. the old houses are oddly shaped like letters that spell out ‘BERLIN’…” Image: Google
Eight expressive word logos: crash, whoosh, shiver, bang, drip, squeeze, roar and wobble
Prompt: “make 8 minimalistic logos, each is an expressive word, and make letters convey a message or sound visually to express the meaning of this word in a dramatic way…” Image: Google
The phrase “How much wood would a woodchuck chuck” spelled out in chopped wood
Prompt: “Create an image showing the phrase ‘How much wood would a woodchuck chuck if a woodchuck could chuck wood’ made out of wood chucked by a woodchuck.” Image: Google

Translate the words, keep the image

With Gemini's multilingual reasoning, the model can "localize and translate your content" while leaving the rest of the image untouched.

Three yellow drink cans with English labels, then the same cans with Korean labels
Prompt: “translate all the English text on the three yellow and blue cans into Korean, while keeping everything else the same” Image: Google

Up to 14 images, up to 5 people

Google says you can blend "up to 14 images" while "maintaining the consistency and resemblance of up to 5 people." The developer guide adds that you can "integrate six high-fidelity shots, or blend as many as fourteen standard inputs into a single, polished ad."

Fourteen fluffy characters squeezed onto a sofa, watching a vintage television
Prompt: “A medium shot of the 14 fluffy characters sitting squeezed together side-by-side on a worn beige fabric sofa and on the floor…” Image: Google
A gown, plants and a chair combined into one cinematic scene with a mannequin
Prompt: “Combine these images into one appropriately arranged cinematic image in 16:9 format and change the dress on the mannequin to the dress in the image” Image: Google

Studio-quality control

Google says the model gives you "control over the physics (lighting, camera, focus, color grading) and composition of the image," with 2K and 4K output for professional production.

A fox in a snowy forest by day, then the same scene at night
Prompt: “Turn this scene into nighttime” Image: Google
A man's portrait relit so that only his eyes and cheekbones catch the light
Prompt: “Generate an image with an intense chiaroscuro effect. The man should retain his original features and expression…” Image: Google
A woman in a flower field, then the same photo with the focus moved to the flowers
Prompt: “Focus on the flowers” Image: Google
A portrait recomposed into new aspect ratios with the subject kept in place
Prompt: “change aspect ratio to 1:1 by reducing background. The character, remains exactly locked in its current position” Image: Google

How to prompt Nano Banana Pro

Google's prompting guide recommends naming each part of the image:

  1. Subject: who or what is in the image, as specifically as possible.
  2. Composition: how the shot is framed, such as a close-up or a low-angle shot.
  3. Action, location and style: what is happening, where, and in what aesthetic.
  4. Text: state exactly what the text says and how it looks, for example "The headline 'URBAN EXPLORER' rendered in bold, white, sans-serif font at the top."
  5. Reference roles: when you upload images, say what each one is for, such as "Use Image A for the character's pose, Image B for the art style."

Specs

Nano Banana Pro
Model IDgemini-3-pro-image
ReleasedNovember 20, 2025; generally available on Gemini Enterprise Agent Platform since May 28, 2026
InputText and images
OutputImages and text
Resolutions1K, 2K, 4K
Aspect ratios15, including 1:8, 8:1, 21:9 and 9:21
Reference images per promptUp to 14
Context window65,536 tokens
Search groundingSupported

Pricing

On the global endpoint of Gemini Enterprise Agent Platform, Google bills image output at $120 per million tokens:

ResolutionOutput tokensApprox. price per image
1K1,120$0.134
2K1,120$0.134
4K2,000$0.24

Text and image input costs $2.00 per million tokens. These are Google's prices as of October 2, 2026.

Nano Banana Pro, 2 or Lite?

Google describes Nano Banana Pro as "optimized for complex, professional use cases," with "the most robust control and advanced reasoning for tasks where accuracy is more important than speed."

If you need…Choose
Maximum accuracy, complex layouts and dense textNano Banana Pro
High quality at Flash speedNano Banana 2
The lowest cost for 1K draftsNano Banana 2 Lite

Frequently asked questions

Is Nano Banana Pro the same as Gemini 3 Pro Image?

Yes. Nano Banana Pro is the product name; gemini-3-pro-image is the model ID developers use.

When was it released?

Google announced it on November 20, 2025. Google Cloud lists the generally available version on Gemini Enterprise Agent Platform from May 28, 2026.

How many reference images can I use?

Up to 14 images per prompt, keeping the resemblance of up to 5 people.

How can I tell an image was made with AI?

Google marks images made with Nano Banana Pro with its invisible SynthID watermark, and the model supports C2PA Content Credentials.

Sources

  1. google-blogIntroducing Nano Banana Pro — by Google
  2. google-blogBuild with Nano Banana Pro, our Gemini 3 Pro Image model — by Google
  3. google-blog7 tips to get the most out of Nano Banana Pro — by Google
  4. google-deepmindGemini 3 Pro Image — by Google DeepMind
  5. google-deepmindGemini 3 Pro Image model card — by Google DeepMind
  6. google-aiGemini 3 Pro Image (Gemini API) — by Google AI for Developers
  7. google-cloudGemini 3 Pro Image (Nano Banana Pro) model card — by Google Cloud
  8. google-cloudGemini Enterprise Agent Platform pricing — by Google Cloud