On November 7th, it was reported that Google officially revealed the origin of the name "Nano Banana" for its AI image model: it was originally just a placeholder idea conceived late at night, but due to its popularity among users, Google simply adopted it.
In the latest episode of the "Made by Google" podcast, product manager David Sharon stated that the model's official name is actually Gemini 2.5 Flash Image. At the time, a product manager named Nina needed users to enter a temporary name for anonymous testing of the model on the LM Arena benchmark website. At 2:30 AM, she had a flash of inspiration and casually wrote "Nano Banana." Unexpectedly, the model quickly became a hit online, with users calling it by this amusing nickname, so Google decided to officially use it.
Sharon revealed that the team even added multiple banana icons within the Gemini app to indicate the model's location, truly "playing with the meme." Regarding the technical highlights, Sharon said that what truly surprised him was the model's accuracy in reproducing human faces. The ability to generate characters that truly resemble the user, rather than being an "AI-generated similar version," is one of the reasons this feature has quickly become so popular.
He also mentioned some globally popular creative trends: such as the "doll" prompts originating from Thailand with only 90 words, Polaroid collage-style photos, and the fun of restoring old family photos. At the same time, he emphasized the security measures for generative AI images—Gemini-generated images all have visible watermarks and invisible SynthID tags to ensure traceability.
For first-time users of "Nano Banana," Sharon suggests starting with the basics. The Gemini app offers ready-made templates, such as the once-popular "doll" feature: upload a photo and see yourself transformed into a new character. Afterward, you can continue to try different outfits, scenes, or eras to explore even more possibilities.
