What is Nano Banana (Gemini image model)?

Nano Banana is Google's name for the Gemini image generation and editing models, not a separate tool from Gemini. The family runs from the original Nano Banana (Gemini 2.5 Flash Image, shut down by Google on October 2, 2026) through Nano Banana Pro (Gemini 3 Pro Image) and Nano Banana 2 (Gemini 3.1 Flash Image) to Nano Banana 2.1, generally available since October 2026.

The name originated on AI image leaderboards where the first model appeared anonymously as 'nano-banana' before Google attribution was confirmed, and the models were widely adopted for product-photo workflows because of their identity consistency, prompt adherence, and ability to faithfully render real products from reference images. Two capabilities matter most for AI UGC: first, the models accept multiple reference images (e.g., a product shot + a persona shot; Nano Banana 2.1 takes up to 14) and compose them into a new scene without losing either subject's identity—solving the historical pain point of 'the model invented its own version of my product'; second, they are responsive to natural-language scene direction, so a single sentence can shift lighting from harsh midday to golden hour, swap a kitchen for a balcony, or change the persona's expression without re-rolling the entire image. Nano Banana 2.1 costs half Nano Banana 2's price per image at 1K and 2K and renders on-image text more reliably. For e-commerce sellers and DTC brands, Nano Banana is the substrate that makes 'upload your product once, render it in any scene' actually work at production quality.