Nano Banana 2

Google's fast image generation and editing model with fal.ai web search grounding, reference images, and resolution controls.

Model Overview

Nano Banana 2 generates and edits images from natural language prompts. It can use optional reference images and web search grounding when the prompt needs current information.

Best At

• Prompt-driven image generation and image editing.
• Using reference images for product, character, or style guidance.
• Producing context-aware images when web search grounding is enabled.
• Fast visual iteration with explicit aspect ratio, resolution, and output format controls.

Limitations / Not Good At

• Web search adds cost and should be used only when current information is needed.
• Highly constrained layouts, exact typography, and brand-perfect edits may need iteration.
• Reference image quality and prompt specificity strongly affect edit quality.

Ideal Use Cases

• Current-event visuals, editorial images, and topical creative assets.
• Product or character variations from one or more reference images.
• Social, marketing, and concept imagery that needs fast iteration.
• Image edits described in natural language.

Input & Output Format

• Input: Required prompt text, optional image references, aspect ratio, resolution, output format, safety tolerance, web search, and seed.
• Output: Generated or edited image assets returned on response.

Performance Notes

• 1K is the default resolution. 2K and 4K increase cost. Web search adds a separate surcharge.