How Nano Banana works: the new Gemini AI model that is in everyone’s mouth

Google has integrated “Nano Banana”, its new AI-powered image editing model, into the Gemini app, and everyone is showcasing their creations on social media. And this is certainly for good reason.

Our new native image generation and editing is state-of-the-art, and ranked #1 in the world. And we’re rolling it out for free to everyone today.

You have the tools. Now go bananas. Ideas & inspiration in the 🧵below. pic.twitter.com/mw7XyG5nes

— Google Gemini App (@GeminiApp) August 26, 2025

Edit images in Gemini without undesired variations

Consistency is one of the greatest challenges in image editing with artificial intelligence, and it appears that Google has successfully overcome this with its latest update. Thanks to its Gemini 2.5 Flash Image model (codenamed “Nano Banana”), you may now upload an image to Gemini, request an edit, and achieve a result with no more variations than those you have specified.

Google launched its native image editing functionality in Gemini in April of this year, and since then it has continued to work to enhance its capabilities and ensure the preservation of similarity between images. “Our latest update is designed so that photos of your friends, family, and even your pets always appear the same, whether you are trying out a 1960s beehive hairstyle or putting a tutu on your chihuahua,” the technology company confirms.

If you also wish to join the “Nano Banana” trend and edit your own images for free, you only need to access Gemini, upload the image you wish to edit, and enter your instructions. Furthermore, once you have your edited image, you may re-upload it and convert it into a short video if you are subscribed to Google AI Pro.

Below are several examples of the editing capabilities of this model and what you can achieve with it:

Change of location and outfit

You can easily modify the background or setting of any image, whether you would like to appear next to a monument, place your pet in a magical world, or test ideas for changing the color of your bedroom walls. Simply write what you wish to edit, and Gemini will handle it.

You can also ask the AI to dress you in different garments than those shown in your reference image. Moreover, this is not limited to clothing alone—try out that hairstyle you have been considering or try a new hair color. All of this is accomplished without causing major alterations to your face, as often happens with other models.

Merge photos

Another particularly interesting feature is the ability to merge elements or characters from two images to create a new one in which they both appear together.

• You may be interested in: Bring your stories to life with the new Gemini Storybook picture books

Combine designs, textures, and styles

You may also upload two different images and ask Gemini to integrate the style, color, texture, or design of an element from one image into the design of the other. Alternatively, you can upload a single reference image and instruct the AI to generate something specific by applying its design. For example, have the wings of a butterfly integrated into the design of a spectacular dress.

Multi-turn editing

Lastly, we wish to highlight the fact that once you have made an initial edit with Gemini, you may continue to use the chatbot to request further modifications to the image without needing to re-upload it, while preserving all previously made changes.

Visible and invisible digital watermark

Google has decided to integrate a visible watermark in the lower right corner of all images generated or edited with Gemini, in order to help users identify this type of content. Additionally, as is the case with its other AI-generated products, these images include SynthID technology, an invisible digital watermark that identifies the content as AI-generated.

Photo: Variations of an image from Depositphotos created with Gemini