Nano Banana (Gemini Image)

Nano Banana is Google DeepMind's family of image generation and editing models built on top of Gemini, released as the visual counterpart to the Gemini text models. The lineup includes Nano Banana Pro (Gemini 3 Pro Image) for studio-quality generation and precise control, Nano Banana 2 (Gemini 3.1 Flash Image) for faster professional-grade output, and Nano Banana 2 Lite for very fast, low-cost generation. All variants accept multimodal input, meaning you can prompt with text, reference images, sketches, or a mixture, and iterate conversationally by asking the model to change lighting, swap objects, extend the canvas, or re-render a scene with a different subject while preserving identity. Because the models inherit Gemini's world knowledge and reasoning, they are noticeably better than typical diffusion models at prompts that require literal accuracy: readable text inside the image, correct diagrams, product mockups with real logos in the right spots, infographics, comic panels with consistent characters across frames, and photo-realistic composites that respect physics and spatial relationships. Typical workflows include marketing hero images, storyboards, product visualisations, character sheets for games and animation, e-commerce catalogue shots, editorial illustrations, UI mockups, and rapid iteration on a base image via follow-up chat turns. The models are available inside the consumer Gemini app, through Google AI Studio for prototyping, via the Gemini API for developers, and inside the Gemini Enterprise Agent Platform and Vertex AI for larger deployments. All outputs are watermarked with SynthID for provenance. It is aimed at designers, marketers, product teams, indie developers, and anyone who wants both generation and conversational, instruction-following editing in a single model rather than stitching separate tools together.