MMOOZZGG.com

Glossary

What is Nano Banana?

In short

Nano Banana is the name of a family of Google models that create and edit images from a text description.

Models and products · Nano Banana

MMOOZZGG.com editorsChecked: 1 min read

For a child

Imagine a magic pencil. You tell it: “Draw a cat in a hat on a bicycle”, and the drawing appears. Then you ask: “Make the hat red”, and the pencil changes only the hat, while the cat stays the same. On every drawing it puts an invisible mark, so everyone knows that a machine drew this picture, not a person.

In plain words

You describe the picture you need in words, and the model draws it. You can upload your own photo and ask it to remove or add something or change the style, and the rest of the picture stays as it was. The model combines several source images into one and keeps a character’s look across different scenes. It writes readable text on images, so it suits menus, diagrams and infographics. For example, a cafe owner takes a photo of a dish and asks the model to turn it into a poster with the name and price.

For business

Marketing teams use such models for banners, product cards, packaging mockups and article illustrations. Every image created gets an invisible SynthID watermark that identifies it as generated: this matters for honest advertising. Access comes through the Gemini API, Google AI Studio and Vertex AI. Russia is not on the list of countries where Google AI Studio and the Gemini API work, or on the list for the Gemini app. Nano Banana has no open weights, so you cannot run it on your own servers. If images are needed inside a closed Russian environment, teams compare open models and the cost of their own server; such options are reviewed in the AI solution map.

For a specialist

Under the name Nano Banana, Google releases several models of the Gemini line. The first, gemini-2.5-flash-image, was introduced on 26 August 2025 as a model for image generation and editing. Then came Nano Banana Pro (gemini-3-pro-image), Nano Banana 2 (gemini-3.1-flash-image), the lighter Nano Banana 2 Lite and Nano Banana 2.1, which the documentation recommends for new projects. Modes: text-to-image, editing an existing image by a text command, combining up to 14 references (the limit depends on the model), keeping a character between frames and style transfer. The models rely on Gemini’s knowledge of the world and draw legible text. All results get a SynthID watermark. Google’s documentation calls the first model outdated and advises moving to the new versions.

Example

A store uploads a product photo on a white background and asks to place the product on a kitchen table in morning light. A minute later a catalog card is ready, and the SynthID mark shows that AI created the scene.

Related terms

Questions and answers

How can you tell a Nano Banana image from a photo? Every image created or changed by these models gets an invisible SynthID watermark. The eye cannot see it, but special Google tools use it to detect that AI made the picture.
Can you use Nano Banana in Russia? Russia is not on Google’s official lists of countries for Google AI Studio, the Gemini API and the Gemini app. The model has no open version for your own servers.

Sources

Not sure whether your company needs this?

AI solutions map
Contact us