All the Features of Google’s Nano Banana 2 Image Generator.
Last year, Google’s Gemini AI took a significant step forward in image generation with the launch of the Nano Banana update—perhaps due to its unusual codename, not just its impressive capabilities —and now the next version has arrived. Nano Banana 2 is another notable update to AI image generation, and it’s currently rolling out to all Gemini users.
Nano Banana 2 is a combination of the original Nano Banana and the Pro version, released a few months later. Essentially, it’s a faster Nano Banana Pro, as noted in Google’s announcement , though Nano Banana Pro will also remain available to users of the Plus, Pro, and Ultra plans for situations where detail and accuracy are more important than speed.
If you’re following the technical names of these models, the Nano Banana is the “Gemini 2.5 Flash Image,” the Nano Banana Pro is the “Gemini 3 Pro Image,” and the Nano Banana 2 is the “Gemini 3.1 Flash Image” (reflecting its improved flash capabilities and speed).
What can Nano Banana 2 do?
The Nano Banana 2 inherits most of the Nano Banana Pro’s features, adding enhanced world awareness (enabling real-time information, such as weather forecasts) and accurate and legible text (often a weakness of earlier AI-based image processing models). Google also emphasizes object consistency, the ability to follow detailed instructions, aspect ratio and resolution control, and high image quality.
While Nano Banana 2 is now available to all users, as you might expect, there are usage limits . You can create 20 images per day without a subscription, 50 images per day for AI Plus subscribers, 100 images per day for AI Pro subscribers, and 1,000 images per day for AI Ultra subscribers. (Google warns that “limits may change frequently” based on demand.)
The Nano Banana 2 will replace the Nano Banana Pro in almost every way you use Gemini, including AI search and Google Lens. As always, generated images will be tagged with Google’s SynthID technology, indicating that they were created using AI.
Understanding exactly what’s changed isn’t easy, but essentially, free users previously had very limited access to the Nano Banana Pro—sometimes just one or two captures per day, depending on overall demand. Thanks to the Nano Banana 2’s improved performance, these users can now create many more photos with the model, which is nearly as good as the Pro (at least until Google adjusts usage limits again).
For paid users, the Nano Banana Pro remains available, although the Nano Banana 2 will be the default. This shows that the Pro is still slightly better (albeit slower), and Google says it’s better suited for “high-precision tasks that require maximum factual accuracy” and where “additional detail” is needed .
Testing the Nano Banana 2
Overall, when you start testing the Nano Banana 2, you might not notice a significant difference compared to the Nano Banana Pro—it’ll just take photos faster. I tested the new model in a variety of shooting conditions, and it’s certainly impressive, even if the results aren’t always perfect.
To get started, select “Create Image” in the Gemini app, and you’ll see a new template feature: you can choose one of the preset templates, like “Gothic Clay” or “Oil Painting,” and then add it to your prompt, or simply enter a prompt from scratch as usual. As before, you can also create a source image (or multiple images) for Gemini to work with using the “+” (plus) button.
If you subscribe to the Gemini AI Plus, AI Pro, or AI Ultra plans, after the Nano Banana 2 has processed your image, you can tap the three dots below it and select “Repeat with Pro” to re-process the same query using the Nano Banana Pro’s extra processing power (though this seems to somehow delete the image template if you selected it).
In one of my tests, I asked Gemini to create a papercraft-style weather forecast for New York City—using image generation, real-time information, and text display—and it performed well, using both the Nano Banana 2 and Nano Banana Pro. The two models produced very different, but equally high-quality results. As expected, the Pro model offered slightly more detail.
I also asked for a test of landscape painting and infographic creation skills, and even commissioned a comic strip illustrating the opening of Charles Dickens’s Bleak House (a topic worth investigating for copyright lawyers). Overall, the results were impressive: the text and graphics were accurate, the style matched the instructions, and there were few errors.
AI-generated images are not yet perfect.
These models aren’t perfect. The texts sometimes contain spelling errors, the image physics are sometimes flawed (AI still can’t accurately recreate the real world), and the graphics still look like templates trained on a huge amount of previous content rather than something truly original.
Comparison of Nano Banana 2 and Nano Banana Pro
Comparing the Nano Banana 2 and Nano Banana Pro directly is difficult, as the template is obviously remade from scratch each time. Therefore, choosing “Remake” in the Pro means you get something original, not just a reworking of the original image. Overall, the Pro seems to surpass the Nano Banana 2 in quality and subtlety, but in some cases I preferred what the Nano Banana 2 offered.
After numerous test runs, it seems the biggest advancement so far is in how Gemini can extract information from the internet (such as weather conditions or specific details) to create images. These models are also getting better at handling different styles and leaving fewer and fewer signs that the images were taken by artificial intelligence. Of course, your mileage on this will vary from person to person.