Nano Banana 2 brings improved image generation features to Gemini for free

Google has just announced Nano Banana 2, which generates more consistent images without compromising speed. It uses the Gemini 3.1 Flash Image model. Google says it relies on Gemini’s perception of the real world and extracts information and images on the web in real-time for more accurate and realistic results.

If you’ve struggled with gibberish, or sometimes the extra letters, for text, Nano Banana 2 is likely to fix it. In addition to creating accurate stylized text on graphics, Nano Banana 2 can also recreate images by translating text into other languages, easing the process of localizing documents or creatives for different regions.

A big upgrade coming with Nano Banana 2 is its ability to maintain consistency across multiple images. Nano Banana 2 can also support consistency for up to five characters and combine up to 14 different objects from input images into a single output, which Nano Banana Pro already supports. Another aspect it borrows from the Pro version is the ability to interpret more complex instructions than before without missing the nuances.

Along with the other changes, Nano Banana 2 also improves image quality and can now create images with up to 4K resolution, surpassing its current cap of 2K resolution. We saw hints about this functionality in Gemini recently, along with the option to choose between 2K and 4K output.

Nano Banana 2 is replacing the previous generation across a range of apps, including Gemini, Google Search, Google AI Studio, paid API, and Google Ads. It’s also appearing in Google’s AI creative studio, Flow, which was enhanced with image generation skills only recently.