Skip to main content
The Generation API exposes three image models — NanoBanana, NanoBanana 2, and NanoBanana Pro — each with two input modes: text-to-image and image-to-image from URLs.

Base URL

All endpoints are served under:

​
How it works

All generation requests are asynchronous. A successful call returns 202 Accepted with a job_id and a status_url. You then poll GET /jobs/{job_id} until the job reaches a terminal state (done or failed).
1

Create a job

POST to a generation endpoint with your prompt and parameters. You receive a job_id.
2

Poll for the result

GET /jobs/{job_id} until status is done or failed.
3

Download the image

When done, read result.image_url — a presigned URL valid for 24 hours.
Prefer push over poll? Register a webhook and we’ll call you when a job finishes.
Prefer a synchronous call that returns the image inline? Use the Gemini-compatible API with Google’s official google-genai SDK — same key, no polling.

​
Models at a glance

​
Next steps

Quickstart

Create your first generation end-to-end with cURL.

Authentication

Get your API key and learn about idempotency.

Models

Compare NanoBanana, NanoBanana 2, and Pro.

Gemini-compatible API

Use the google-genai SDK against BananaHub.

API Reference

Full request and response schemas.