Nano Banana

In August 2025, Google's Gemini team submitted a new image model to LMArena (a crowd-sourced AI evaluation tool that shows models anonymously so users can compare outputs without brand bias) under the codename 'Nano Banana', a name chosen at 2 AM by a product manager who combined two personal nicknames. The model topped the image editing leaderboard. When Google revealed its identity two weeks later, the name had already stuck. Google leaned into it, adding banana emoji to their AI Studio and Gemini app interfaces.

Technically, Nano Banana refers to a line of image generation and editing models built on Gemini's multimodal architecture. Unlike standalone text-to-image models, these were designed from the ground up to understand text and images together, enabling conversational editing, subject consistency across multiple revisions, multi-image compositing, and targeted edits using natural language. The first version (Gemini 2.5 Flash Image) helped attract over 10 million new Gemini app users and facilitated 200 million image edits within weeks of launch. It also went viral for photo editing trends like the 3D figurine style.

The family has grown: Nano Banana Pro (Gemini 3 Pro Image) is the studio-quality tier, and Nano Banana 2 (Gemini 3.1 Flash Image, rolled out February 2026) brings faster generation and better text rendering. The models are available in the Gemini app, Google AI Studio, through the Gemini API, and in Adobe Firefly as a partner model. If you see someone talking about 'Google's banana model' or just 'Nano Banana', this is what they mean.