Nano Banana API: pricing and image generation

Choose a Nano Banana model

Choose a model for your quality target and budget: compare per-image rates below, then test the result on your own task. For larger jobs, use the queue and budget guide or the ComfyUI workflow.

Nano Banana is a family name. Nano Banana 2 uses gemini-3.1-flash-image, Nano Banana Pro uses gemini-3-pro-image, and the original model uses gemini-2.5-flash-image. Use the exact ID available to your key, not a marketing name.

Current Nano Banana API prices

Image prices are not the full request cost. Input text, reference images and other billable operations may add charges. A published model does not prove access for your key: check GET /v1/models.

Get an API key and generate an image

Create a key in your API429 account, check your balance and confirm model access. Set API429_API_KEY as a server-side environment variable. This example makes a paid request.

curl https://gateway.api429.com/v1/images/generations \
  -H "Authorization: Bearer $API429_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gemini-3.1-flash-image","prompt":"Product photograph of a ceramic mug on a light table, no text","n":1,"size":"1024x1024","response_format":"b64_json"}'

Decode the returned base64 image and save it in your application storage. Keep your key off the browser. See the API documentation for the request contract and response handling.

Nano Banana in ComfyUI: download a ready workflow

Use the API429 custom node to generate images without downloading model weights. The package includes text-to-image, one-reference editing, a sequential prompt queue and a public-price lookup. Results connect to the standard ComfyUI IMAGE output and Save Image.

Install the API429 node and follow the English setup guide · Download the Nano Banana workflow · Comfy Registry listing

1. Install the node into your ComfyUI custom_nodes folder and its dependencies in the ComfyUI Python environment.
2. Set API429_API_KEY in the server environment and restart ComfyUI. The workflow contains no API key.
3. Import the JSON, choose the model and prompt, set a unique run_id and enable paid generation when ready.

The live test covers this text-to-image example. FLUX.2 Pro, GPT Image, editing and queue scenarios have local tests only. API429 is an independent custom-node integration, with a separate account and billing from Comfy partner nodes.

Run a queue of Nano Banana images in ComfyUI

Download the image queue workflow (v0.2.0) to process one prompt per line, with up to 100 prompts per run. Choose the model and resolution, set a batch ID, and connect the output to Save Image. Requests run sequentially; the prompt limit is not a dollar spending cap.

If a later prompt fails, keep the same batch ID and unchanged parameters to reuse saved responses. Save Image receives the output when the whole queue completes; partial results remain in the local state store. See the workflow documentation and recovery instructions before starting a large run.

Plan a bulk image-generation budget

Calculate the image component as the number of billable output images multiplied by the rate for the selected model and resolution. Add other billable operations and account for results you reject during review. Divide the total actual spend by the number of accepted images to compare usable output cost.

A client-side queue of individual requests is different from a provider Batch API. Confirm model support, input format, completion time and pricing before using a batch route. A Google Batch discount does not automatically apply to API429.

Generation, editing and application use

Use text-to-image generation for illustrations, creative drafts and product-media concepts. Editing an existing product photo is a separate operation: check support for reference images and /v1/images/edits. Review product details, faces and text before publishing results.

Store each job and its result. After a timeout, reconcile the original request before retrying: an unanswered request may still have completed and been billed. Limit concurrency according to your available limits and measured results.

Frequently asked questions

Is the API free or unlimited?

API429 charges for usage. Access depends on balance, model availability and account limits; an API key is not an unlimited generation subscription.

Is API429 the official Google API?

No. API429 is an independent gateway with its own endpoint, key and billing. Check the supported request format in our documentation before migrating a Google integration.

Where can I check pricing before a large job?

Use the current catalog and your account. Fix the model and resolution, run a small pilot and inspect actual charges before scaling.