Who Created Nano Banana? Sundar Pichai Confirms Google Behind Viral AI Image Tool Gemini 2.5 Flash Image

Google has finally revealed the source of the viral image generation tool “Nano Banana.” On Sunday night, the company’s chief executive, Sundar Pichai, tweeted three banana emojis, further intensifying speculation that Google was involved in the eccentric project.

Its name, Nano Banana, is actually the codename for its latest AI-driven photo editing model, Gemini 2.5 Flash Image, the company disclosed hours later.

The newly launched AI tool comes as part of the Gemini app, and it promises context-aware, high-quality photo editing. Gemini 2.5 Flash Image is not only an AI photo editor but also keeps the main object of the photo, like a person, animal, or product, unchanged even after multiple edits.

Sundar Pichai Dog Edits

To celebrate the launch, Pichai demonstrated the model’s capabilities on X (formerly Twitter) using his dog, Jeffree, as the subject. In the playful posts that celebrated International Dog Day, Jeffree was turned into a Hawaiian surfer, a flamboyant cowboy, a larger-than-life superhero, and a larger-than-life chef—and all of them still looked like Jeffree.

“Top of @lmarena’s image edit leaderboard—especially good at retaining likeness in a wide range of contexts,” Pichai wrote, noting or joking that Jeffree could have an affinity for the couch as opposed to stunts.

How does the Gemini 2.5 Flash Image work?

The new AI editor builds on previous image tools in Gemini. One of the main challenges in AI photo editing has been keeping the subject accurate. Often, pets and people end up distorted or unrecognizable. Nano Banana provides solutions to this problem by making subjects stay consistent across edits.

Users can upload a photo, write a short prompt, and the AI will do the transformations. In addition to simple editing, the model also supports multi-turn editing, and users can generate scenes without manually specifying everything at once.

Another is design remixing, where patterns or textures from one image are copied into another. For instance, they can turn butterfly wings into fabric patterns or reimagine rain boots with the texture of a petal.

The program also provides an option of photo blending, so you can combine several pictures for an elegant twin effect (such as face morph). And edited photos can be re-fed into Gemini to create animated videos.

Available Worldwide

According to Google, the Gemini 2.5 Flash Image is now live globally for free and premium Gemini users. Each AI-generated photo will also be wet-stamped (visible and invisible) in the SynthID so that AI-created content can be effectively and visibly labeled.

For businesses and developers, the new tool is available through the Gemini API, Google AI Studio, and Vertex AI. Pricing has been established at $30 per million output tokens, while individual image generations cost around $0.039 each.