Nano Banana 2 Lite AI Image Generator
Generate and revise 1K images quickly with Nano Banana 2 Lite, powered by Google DeepMind Gemini 3.1 Flash-Lite Image. It is built for rapid prompt iteration, reference-guided edits, concept exploration, and everyday content production.
This workflow is not currently available.
Fast 1K Image Generation
Move from prompt to usable 1K image quickly, making the Lite model a practical choice for thumbnails, social posts, mood boards, rough campaign directions, and teams that need many creative iterations.

Prompt-Based Image Editing
Upload source images and explain the change in everyday language. Revise objects, styling, setting, color, or composition without building a complex mask-based editing workflow.

Up to Ten Reference Images
Provide as many as ten references to communicate a subject, product, style, or layout. Reference-guided input helps keep visual intent more consistent across a sequence of fast experiments.

Efficient Creative Iteration
Use the model for exploration where speed and cost matter more than maximum output resolution. Generate alternatives, compare directions, and refine the strongest result with a clearer follow-up instruction.

How to Use Nano Banana 2 Lite
Create a image with a focused brief and the source media required by the selected workflow.
Select a creation mode
Choose text-to-image for a new concept or image editing when you already have source material.
Add prompt and references
State the intended subject, style, composition, and change. Upload up to ten relevant images when visual guidance is helpful.
Iterate at Lite speed
Generate a 1K result, compare variations, refine the prompt, and download the image that best matches your brief.
Models Related to Nano Banana 2 Lite
Nano Banana 2 Lite Frequently Asked Questions
Nano Banana 2 Lite is the fast image-generation and editing option based on Google DeepMind Gemini 3.1 Flash-Lite Image. It targets efficient 1K creation and rapid prompt iteration.
Create with Nano Banana 2 Lite
Explore ideas, edit references, and produce 1K images with a fast iteration loop.
