Images

The image generation endpoint produces images from natural language prompts. It is OpenAI-compatible, so the openai SDK's images.generate call works without modification — just swap the base URL. It maps to POST /v1/images/generations.

Generate an image#

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://opendunes.com/v1",
  apiKey: process.env.OPENDUNES_API_KEY,
});

const resp = await client.images.generate({
  model: "google/gemini-2.5-flash-image",
  prompt: "An aerial view of the Sahara dunes at golden hour, photorealistic",
  size: "1024x1024",
  n: 1,
});

console.log(resp.data?.[0]?.b64_json?.slice(0, 32)); // the image itself, base64-encoded

Each image comes back in the response: its bytes base64-encoded in data[].b64_json, its type in media_type, and the id of the copy kept in Files in file_id. A kept image never expires. If the provider returns no image, the call fails with 502 no_image and you are not charged.

Editing a picture goes through the same route: send the start picture in input_references — see Image Generation. The SDK's images.edit and images.createVariation call routes OpenDunes does not carry.

Pricing#

Image generation is billed per output image — n images cost n × the model's per-image rate. Rates are shown in the model catalog. See the image API reference.