# Nano Banana Nano Banana is the name for the [[Gemini]] family's native image generation and editing models from [[Google DeepMind]]. It began as an internal codename and stuck — "Nano Banana" is now the public brand for Gemini image generation. By Google I/O 2026, Google reported over 50 billion images generated with these models. ## The variants - **Nano Banana** — the original Gemini Image model; conversational image generation from text, images, or both - **Nano Banana 2** (Gemini 3.1 Flash Image) — pairs the quality of Nano Banana Pro with Flash-tier speed; pulls real-time information and images from web search to render specific subjects more accurately - **Nano Banana Pro** (Gemini 3 Pro Image) — the state-of-the-art tier, built on Gemini 3 Pro; uses Gemini's reasoning and world knowledge to render information-dense images ## How it works Image generation is conversational and iterative — describe what you want, then refine by talking to the model. It is the same loop [[Gemini Omni]] brings to video. Every output carries both an invisible SynthID watermark and a visible mark to signal it is AI-generated. ## Why it matters - Nano Banana is Google's entry in the image-generation race; the 50-billion-image figure shows it is a high-volume consumer surface, not a side project - The naming now flows outward: [[Gemini Omni]] is pitched as "Nano Banana, but for video," which makes Nano Banana the reference point for Gemini's whole generative-media line ## References - Gemini Image (Nano Banana): https://deepmind.google/models/gemini-image/ - Nano Banana Pro: https://blog.google/innovation-and-ai/products/nano-banana-pro/ - Nano Banana 2: https://blog.google/innovation-and-ai/technology/ai/nano-banana-2/ ## Related - [[Gemini]] - [[Google DeepMind]] - [[Gemini 3]] - [[Gemini Omni]] - [[Gemini 3.5 Flash]]