Google has released its highly-anticipated image generation and editing model, the "Nano Banana Pro" on Thursday.
It is built on top of the company's new large-language model Gemini 3.0.
Nano Banana in its earlier avatar went viral on social media. While it was no match for ChatGPT's Studio Ghibli sensation, many users did enjoy the speed at which Nano Banana generated images and how good it was at editing or retouching the same image again, without introducing unwarranted changes that many other LLMs were struggling with.
While I have not had a lot of time to test the model launched roughly an hour ago yet, it seems to be impressive and up a notch from its prior avatar.
I tried to get it to generate an infographic around progression of the Gemini line of models and I was surprised to see that it first ran a web search to fetch accurate information and then generated the image with that text.
It is hard to imagine now how impossible it felt just a year or so ago to even have a proper word written in an image generated by AI.
Alongside the model, Google also announced it would watermark all images generated by Nano Banana, with plans to bring this to other media, and more importantly, users would be able to upload any image to test if it was generated with the company's AI!
This would be a vital fact-checking tool at a time AI misinformation is running rampant.