Google Launches Nano Banana Pro AI Image Generator Powered By Gemini 3 Pro

This year, Google's AI image generation team managed to pull off a trick unseen from rivals. The Nano Banana image generation model (Gemini 2.5 Flash Image) went viral not once, but twice, during pre-release testing online. It happened in August with Nano Banana, which was just a placeholder until Google embraced it as an official product name. Then, Nano Banana 2 (Nano Banana Pro) went viral a few weeks ago, stunning testers with new image generation improvements beyond what Google had achieved months earlier. Rumors that followed said Google would unveil Gemini 3 Pro and Nano Banana 2 at the same time, but that wasn't exactly the case. Gemini 3 Pro dropped on Tuesday, and Google waited until Thursday to release Nano Banana Pro (Gemini 3 Pro Image), its newest state-of-the-art model.

Interestingly, Google unveiled a new Generative UI experience for Gemini 3, where the AI creates immersive visuals for prompts that demand a visual experience. Google said Generative UI uses an image generator to create the visual experiences, without disclosing its identity. The arrival of Nano Banana Pro makes even more sense in that context. Nano Banana Pro is built on Gemini Pro tech, using its "state-of-the-art reasoning and real-world knowledge to visualize information better than ever before."

Google's announcement on Thursday confirms some of the Nano Banana Pro features that leaked last week. The new image generation model will offer even better character consistency and advanced text rendering than its predecessor (see the following image), which already supports realistic image outputs and text support.

Nano Banana Pro can blend elements better than the previous model. It can use up to 14 images and maintain character consistency for up to 5 people (see samples above). The new AI image generation model supports localized editing features, like changing focus, applying "sophisticated" color grading, changing lighting, and even applying bokeh effects to images. Another big upgrade concerns the resolution. Nano Banana Pro lets users create 2K and 4K images, a significant upgrade over Nano Banana's 1024 x 1024 resolution. In addition, Nano Banana Pro supports more aspect ratios than its predecessors, which will help creators come up with AI images for different uses.

Images generated with Nano Banana will continue to be protected with the invisible SynthID digital watermark. The Gemini app will be able to identify Nano Banana creations made with Google's AI image generators. Even better, images generated by Gemini Free and Gemini AI Pro users will feature visible watermarks (the Gemini sparkle) to help anyone identify the origin of the photos. Professionals who subscribe to the Ultra plan will not have to deal with the visible watermarks.

The most obvious place where you can use Nano Banana Pro is the Gemini app, where it'll be available via the Thinking model when you select the "Create Images" option. Users on the free Gemini plan get a limited quota of free Nano Banana Pro images, after which they'll revert to the Nano Banana model. Google AI Plus, Pro, and Ultra have higher quotas. Nano Banana Pro is also available in AI Mode for U.S. users. NotebookLM users worldwide will also have access to the new image models. Google AI Ultra subscribers can also use Nano Banana Pro in the Flow AI filmmaking tool.

Google also said that Nano Banana Pro will be available in Google Ads, Workspaces, Google Slides, and Vids. Developers can use it in Google AI Studio, Google Antigravity, and via the Gemini API. Finally, Nano Banana Pro will probably roll out to third-party services that support Google's AI image generation tools, but Google didn't mention any partnerships in the announcement.