What Is Nano Banana? Google’s AI Image Model Explained

Nano Banana is the nickname for Google’s image generation and editing models inside Gemini. The first one, Gemini 2.5 Flash Image, went public in August 2025 after testing anonymously under the name "nano-banana". The name stuck, and Google now uses it for the whole family: Nano Banana, Nano Banana Pro, Nano Banana 2, Nano Banana 2 Lite and Nano Banana 2.1.

You type what you want, or upload a photo and describe the change, and the model returns an image. What made it popular is editing: you can say "put a hat on the dog" or "change the background to a beach at sunset" and it changes only that, keeping faces and products recognizable.

Why is it called Nano Banana?

Before launch, Google tested the model on LMArena, a site where people compare two unnamed models side by side and vote for the better image. Using the real name would have given the game away, so the team needed a placeholder.

According to Google’s own account, product manager Naina Raisinghani picked it at about 2:30 a.m. Some of her friends call her "Naina Banana", others call her "Nano" because she is short and likes computers. She combined the two. It also fit because the model was a fast, small "Flash" model.

The anonymous "nano-banana" model kept winning votes, people started searching for it by that name, and Google kept the nickname after launch.

Who made Nano Banana?

Google DeepMind built it, and Google owns it. It is part of the Gemini model family, so "Gemini Nano Banana" and "Google Nano Banana" refer to the same thing. It is not a separate company or app.

The Nano Banana models

The "2" means a newer generation, while "Pro" means a higher tier. That is why Nano Banana 2 is newer than Nano Banana Pro but Pro is still the heavier model. If you are deciding between them, see Nano Banana 2 vs Nano Banana Pro. Google has not announced a Nano Banana 3 yet; Nano Banana 3: what we know tracks the news.

What Nano Banana does well

• Editing by instruction. Describe a change in plain words and it edits only that part of the photo.
• Keeping people and products consistent. The same face or product survives across many edits and new scenes, which earlier image models struggled with.
• Combining images. Newer versions accept many reference images at once, for example a person, an outfit and a location, and blend them into one shot.
• Text inside images. Nano Banana 2 and Pro spell posters, labels and infographics correctly most of the time. The original model was weaker here.
• Speed. The Flash models return an image in seconds, so trying ten variations is quick.

Limits worth knowing

Nano Banana makes still images only. It does not create video or animated GIFs (see Can Nano Banana make videos?). It does not make 3D files either, though a render can feed an image-to-3D tool (see Can Nano Banana make 3D models?). Google adds an invisible SynthID watermark to every image, and the Gemini app also shows a small visible mark by default. Safety filters block some requests, including explicit content and many edits of real people.

Where can you use Nano Banana?

• The Gemini app and gemini.google.com. Free with daily limits, with higher limits on Google AI plans. See Is Nano Banana free? for the details.
• Google AI Studio and the Gemini API. For developers, billed per image. The Nano Banana API guide lists the model IDs and prices.
• Other apps. Photoshop, Canva and many creative tools offer it as one of their models. Where to use Nano Banana compares them on access and price.
• Nano Banana AI (this site). An independent studio, not affiliated with Google, where you can use Nano Banana, Nano Banana 2 Lite, Nano Banana 2 and Nano Banana Pro next to models like GPT Image 2 on one credit balance, without a Google account.

If you want ideas to try first, browse the Nano Banana prompts library.