Also known as “Nano Banana”

In community discussions and external references, Google Gemini 2.5 Flash – Image Preview is often referred to as “Nano Banana”. We use the official model name in this documentation and mention the alias for clarity.

​
Overview

Google Gemini 2.5 Flash – Image Preview focuses on creative precision and reliability. Generate accurate, production‑ready images in a single attempt with Gemini 2.5 Flash, while preserving subject identity and visual coherence. Despite the model name, outputs are solid and ready for production use. Combine your prompt with up to 3 image references to achieve consistent product placement, character continuity, and style alignment. Use webhooks to retrieve results asynchronously in production pipelines.

​
Key capabilities

  • Single‑attempt accurate, production‑ready images for complex transformations
  • Consistent products, characters, and styles across images
  • Multi‑reference prompting (up to 3 image references)
  • Async-friendly: task IDs and webhooks for reliable retrieval
  • Cost‑effective for large‑scale variant testing

​
Use cases

  • Product placement and product + character compositions
  • Character continuity across scenes and expression changes
  • Style alignment and brand consistency for campaigns
  • Ads with text and social creatives at scale
  • Multi‑reference generation and multi‑shot scenes

POST /v1/ai/gemini-2.5-flash-image-preview

Create a Gemini 2.5 Flash generation task

GET /v1/ai/gemini-2-5-flash-image-preview/{task-id}

Get the status of a Gemini 2.5 Flash task

GET /v1/ai/gemini-2-5-flash-image-preview

List all Gemini 2.5 Flash tasks and details

​
Frequently Asked Questions

Yes. “Nano Banana” is a commonly used alias for Google Gemini 2.5 Flash – Image Preview. This API uses the official model name, and all outputs are production‑ready.
Yes. Despite the model name “Image Preview”, outputs are full-quality and ready for production. Use an upscaler only if you need extreme resolutions or larger formats.
Yes. We recommend using webhooks to retrieve results reliably in the background, especially when integrating previews into automated pipelines.
Absolutely. Fast generation and low cost make it ideal for testing multiple prompt variants before committing to high-resolution renders.