Google upgraded Gemini's image-generating chops with Nano Banana and Nano Banana Pro, two new text-to-image AI models, last year. Fun names aside, these models are known as Gemini 2.5 Flash Image and Gemini 3 Pro Image. Google put its best image-generation models in users' hands by making Nano Banana Pro available for three images per day, and consumers are clearly loving it.
Google's vice president of Gemini and Google Labs, Josh Woodward, took to X (formerly Twitter) to announce that users have generated over a billion images with Nano Banana Pro. Woodward noted that this model was only available for 53 days when the major milestone was crossed, signaling that there's significant demand for Nano Banana Pro.
Since free users only get three Nano Banana Pro generations or edits per day, crossing this threshold suggests Gemini users are trying out and returning to the image-generation model. It could also indicate that paid users are taking advantage of Nano Banana Pro access, as Google AI Pro subscribers get 100 images daily and Ultra subscribers get 1,000.
Crossed 1 billion images Nano Banana Pro images in @GeminiApp! The pro community is moving fast. This model has been out for 53 days.Come for the potassium, stay for more. :) pic.twitter.com/DvPoitI3i9January 12, 2026
Nano Banana Pro stands out due to its command of hyperrealism and handling of text. Although text generation is a typical pain point for AI image generators, Nano Banana Pro crosses this hurdle easily with the ability to write text in a variety of languages. There are applications beyond basic image edits or cartoon-style generations, such as creating presentation slides or educational charts.
Users can try out detailed prompts that tell Nano Banana Pro exactly what to do, and then watch the magic happen. Since this is a thinking model, you can trace Gemini's steps as it worked on how to tackle the Nano Banana Pro request. In the example below, a long prompt was used to create an isometric miniature representation of a city.
You'll notice that not only does the output contain legible text using personalized typography, but it also looks clean and accurately matches the prompt instructions.
Nano Banana Pro's image-generation performance seems to be driving more people to use the Gemini app, and it isn't the first time. Back in September, Woodward announced the standard Nano Banana model drove 10 million users to start using Gemini. At the time, the Gemini VP added Nano Banana edited more than 200 million images in just over a week of availability.
Now, Nano Banana Pro is following in Nano Banana's footsteps, marking an even bigger milestone with over a billion edits and generations. This is impressive in and of itself, but it's even better for Google considering the wider context of AI image editing today. As competitors find themselves in scandal, Gemini's Nano Banana is emerging as a fan-favorite image generator that is squeaky clean.
Google's noncontroversial image generator is standing out
Generative AI has been used for nefarious purposes for years, but the average person and global governments are finally understanding the scope of AI image generations and deepfakes, thanks to Grok, Elon Musk's AI chatbot. Grok powers image generation in the X (formerly Twitter) app via xAI and makes it easy to create and edit images on X, sometimes of others and of an explicit nature, by simply replying to a post and tagging the AI bot.
Grok's behavior is nothing new, as the technology has been available on underground sites before. However, it has never been this easy to generate and edit images of others without their consent. It's all happening in public, making Grok's guardrails (or lack of them) and generative AI at large a controversial issue. In response, Grok's image generation features were restricted to paid subscribers.
Things began coming to a head when Malaysia and Indonesia became the first countries to block Grok access due to its unrestricted AI image generation features, as reported by the Associated Press.
By comparison, Google's Nano Banana Pro-powered image generation features are decidedly noncontroversial, and that's beginning to be valuable. The company's guardrails for Gemini image generation, once criticized as being too restrictive, now stand out as potentially the gold standard for reigning in AI bots. Its policy guidelines for the Gemini app prevent Gemini from generating outputs containing violence, gore, explicit material, threats to child safety, harassment, and more.
Beyond including the necessary training limits, Google is showing off how Nano Banana Pro can benefit users by being a workplace and learning tool. It can generate custom explainers in Google Search's AI Mode, create charts or infographics in NotebookLM, or diagrams in Google Slides, for example. Nano Banana Pro is also available in Google Vids and Flow.
Crucially, Google watermarks Nano Banana and Nano Banana Pro generations and edits. Anyone can check if an image (or video, for that matter) was generated with Google AI by uploading it to Gemini and asking.
Google is winning in the AI image generation space on merit, offering a tool consumers want to use, as evidenced by Nano Banana Pro generating over a billion images in under two months. Perhaps more important is that Google isn't compromising principles to win the AI race. Google's image-generation models aim to be safe and transparent at a time when competitors look to be doing the exact opposite.
