The AI model nano-banana, which recently appeared on the LMArena website, has attracted the attention of users with its exceptional image generation abilities. It is not known which company created this model, but the community has some guesses.
The AI model nano-banana, which recently appeared on the LMArena website, has attracted the attention of users with its exceptional image generation abilities. It is not known which company created this model, but the community has some guesses.
LMArena has a feature that allows you to set up a competition between two randomly selected models — that’s where the «nano-banana» model came into being. When it appeared, users started noting how well it generated images, Business Insider reports .
«This is extremely good. Photoshop could end this decade,» wrote one user nano-banana.
AI enthusiasts are trying to figure out who’s behind this model, and most people are now agreeing that it’s Google, in part because the company started hinting at something related to images earlier this month.
Google has not yet said it is the developer of the model, and company representatives have not commented on requests for comment. However, this week, Google’s AI Studio product manager Logan Kilpatrick posted a banana emoji on Twitter. And Naina Risingani, a product manager at Google DeepMind, posted an image that looks similar to Italian artist Maurizio Cattelan’s 2019 work «Banana Taped to the Wall.»
🍌
— Logan Kilpatrick (@OfficialLoganK) August 19, 2025
The use of the word «nano» may indicate that this is a model that can work locally on the device, and Google has previously called its smaller models «nano.» The model may be officially introduced as early as July 20 at a major event dedicated to the company’s new gadgets.
Just two years ago, AI image generators were having trouble with bananas. Back then, scientist Daniel Hook spent weeks bombarding the Midjourney model with queries like «a single banana casting a shadow on a gray background,» but the AI returned an image with two bananas every time. He published an entire paper on the AI’s shortcomings called «The Lone Banana Problem.»
