Google released model for working with images Nano Banana 2.1

The company Google presented an updated model for creating and editing images Nano Banana 2.1 based on Gemini 3.6 Flash.

The new version replaced Nano Banana 2 (Gemini 3.1 Flash Image). The system supports image generation in resolutions 1K, 2K and 4K (1K is set by default) and panoramic formats with proportions up to 1:8 and 8:1 without segmentation artifacts at the joints.

The model has learned to combine up to 14 reference shots simultaneously, maintaining a single appearance of up to four characters and accuracy of up to ten objects in a series of images. The system added a link to web search and Google Images, three customizable levels of reasoning (minimum, medium and high), as well as improved text display and infographic layout.

According to Google, the cost of generating 1K and 2K images in the Gemini API has decreased by two times. Access to the model is open in Gemini, Google AI Studio, the official API, Flow, Stitch, Google Ads, Gemini Enterprise Platform and in the search AI mode.