Key Highlights:
- Google launched Nano Banana Pro, a Gemini 3 Pro–based model that offers sharper, more accurate, and more controllable image generation.
- The model can tap into real-time Search data to create hyper-realistic, educational, and context-rich visuals.
- It’s rolling out across Gemini apps, APIs, and developer tools with higher-quality tiers starting at $0.139 for 1080p/2K outputs.
It wouldn’t be wrong to say that Google has mastered image generation segment with the launch of Nano Banana model earlier this year. Well, the company is building onto that success and moving further with its AI push with the announcement of Nano Banana Pro. It’s a native image model built on the recently release Gemini 3 Pro model. To catch you up, Google recently released the Gemini 3 model which has been praised by users around the globe for its for its performance.
Nano Banana Pro generates accurate images that are contextually rich
Well, Google is carrying over that performance to the Nano Banana Pro, which does a great job when it comes to higher-fidelity image generation and editing. The new model not only focuses on delivering visually appearing images but also makes sure that they are pretty accurate and contextually rich.
Not to forget the new model can also utilize the Search to refer real-time data present online to make a hyper-realistic image. Think of Nano Banana Pro creating images that can used in the educational setting. You can even ask it to look for a recipe online and generate flash cards based on that information.
Google also mentions that Nano Banana Pro gives professional more control over the generated image. You can control different aspects, such as camera angle, lighting, depth of field, color-grading and more. Speaking of field of depth, Meta’s new SAM 3D model also has this feature that it uses to create highly-detailed 3D models of different objects as well as humans.
Nano Banana Pro can render images up to 4K quality
Coming back to Nano Banana Pro, Google says the new model can also generate images in higher quality, though the process might feel sluggish and is costlier than the original model, which costs $0.039 per 1024px image. Meanwhile, the new model requires you to pay $0.139 per 1080p or 2K image. If you want ever higher quality, you can always go for 4K but it will set you back by $0.24. The latest model can use six different high-fidelity shots and up to 14 objects within an image, while maintaining consistency and resemblance of up to five people. Google also says that there’s even a demo app now to test these capabilities.
As of now, the new Nano Banana Pro model is rolling out across all of Google’s existing AI tools. Starting today, Gemini app will use the new model to generate any images by default. If you are a non-paying user, the number of image you can generate will be limited. Once you hit the set cap, the Nano Banana will take over and you can continue to generate images using the original model.
Availability
The Mountain View giant says that Google AI Plus, Pro, and Ultra subscribers will have higher image generation limit without detailing much. These subscribers will also have access to Nano Banana Pro in NotebookLM, which recently rolled out customizable video overviews to everyone. For AI Pro and Ultra subscribers in the US, the new model will also be available in Search via AI mode. Meanwhile Ultra subscribers can access the video tool, Flow, as well. If you are a developer, you can access Nano Banana Pro via the Gemini API and Google AI Studio and the new IDE, Antigravity.
