September 5 News.GoogleThe latest AI experiment, "Nano Banana," burst onto the scene last week, and then Google announced that it'd be launching the newest version of the program at Gemini Gemini 2.5 Flash Image integration is now online.
Last night, Josh Woodward, VP of Google Labs, revealed on X that since the feature went live, more than 200 million image edits have been made, driving more than 10 million new users to try the Gemini app. He described the product's popularity as internal "severe TPU overload and non-stop SRE alerts."
Nano Banana first appeared on the beta platform as a web sensation after users found that it was able to avoid the "Valley of Terror effect" common to other AI image generators, and that it performed well on both figures and real people, was easy to use, and outperformed GPT-4o in several ways.
In addition, unlike some competing products that are prone to facial distortion or loss of character resemblance after multiple edits, the model maintains consistent character traits through repeated modifications. For example, users can change the color of a room, add clothing to a pet, or replace a character's look multiple times while the image still maintains clear personal recognition.
This feature made the Nano Banana hit the LMArena charts even before its official release, and it's been a hit ever since. 1AI found out that Gemini users can currently edit 100 images per day for free, and up to 1,000 images per day for paid subscribers.
According to analysts, Google's AI strategy is clear: attract users with intuitive and high-quality tools, and then bring them into the Gemini ecosystem. Unlike most AI editors that remain in the demo stage, Gemini 2.5 Flash Image has already shown real-world value.