Image generation

POST/v1/images/generations

Generate or edit images through a unified OpenAI Images-compatible endpoint.

The gateway always requires model and prompt. Availability and accepted values for other fields, as well as the image response form, depend on the selected model and Provider route.

Read the complete image guide for references, edits, Base64, and backoff polling

Core request fields

Field	Type	Required	Description
`model`	string	Yes	An image model ID returned by /v1/models.
`prompt`	string	Yes	The image generation or editing prompt.
`image_url`	string	No	A reference image URL when supported by the model.
`image_urls`	string[]	No	Multiple reference image URLs when supported by the model.
`bbox_list`	integer[][][]	No	Reference image regions when supported by the model.
`enable_sequential`	boolean	No	Enables sequential generation when supported.
`thinking_mode`	boolean	No	Enables model thinking mode when supported.
`n`	integer	No	Requested image count; it must be greater than 0.
`size`	string	No	Requested dimensions; accepted values depend on the model.
`resolution`	string	No	Requested resolution; accepted values depend on the model.
`quality`	string	No	Requested quality tier; accepted values depend on the model.
`background`	string	No	Background mode when supported by the model.
`mask_url`	string	No	Mask URL for local edits when supported by the model.
`style`	string	No	A style hint when supported by the model.
`watermark`	boolean	No	Requests a watermark when supported by the Provider.
`response_format`	string	No	Requests URL, Base64, or another supported response form.
`seed`	integer	No	A random seed when supported by the Provider.
`guidance_scale`	number	No	Prompt guidance strength; range depends on the model.
`stream`	boolean	No	Streaming support depends on the route and is unavailable on some routes.
`output_format`	string	No	Output file format when supported by the model.
`sequential_image_generation`	string	No	Sequential generation mode when supported.
`sequential_image_generation_options`	object	No	Model-specific sequential generation options.
`tools`	object[]	No	Model tools forwarded only by supporting routes.
`optimize_prompt_options`	object	No	Model-specific prompt optimization options.
`user`	string	No	A caller-defined user identifier.
`metadata`	object	No	Caller-defined metadata.

Request examples

curl https://api.noxgate.net/v1/images/generations \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"YOUR_MODEL_ID","prompt":"A quiet library at sunrise","n":1,"size":"1024x1024"}'

import requests

response = requests.post(
    "https://api.noxgate.net/v1/images/generations",
    headers={"Authorization": "Bearer sk-your-api-key"},
    json={"model": "YOUR_MODEL_ID", "prompt": "A quiet library at sunrise", "n": 1},
)
print(response.json())

const response = await fetch("https://api.noxgate.net/v1/images/generations", {
  method: "POST",
  headers: { Authorization: "Bearer sk-your-api-key", "Content-Type": "application/json" },
  body: JSON.stringify({ model: "YOUR_MODEL_ID", prompt: "A quiet library at sunrise", n: 1 }),
});
console.log(await response.json());

Successful response

{
  "created": 1750000000,
  "data": [{ "url": "https://provider.example/generated-image.png", "size": "1024x1024" }],
  "usage": { "generated_images": 1 }
}

Retrieve an asynchronous task

GET/v1/images/generations/:task_id

When creation returns status and id without usable data, poll this endpoint by task ID. Treat succeeded, failed, and canceled as terminal states and enforce a maximum wait time.

curl "https://api.noxgate.net/v1/images/generations/image_task_example" \
  -H "Authorization: Bearer sk-your-api-key"

Usage notes

Do not assume every image model accepts reference images, the same sizes, or the same optional fields.
Each response data item may contain url or b64_json; handle the field actually returned.
Billing follows the selected model configuration and actual usage; this guide does not fix an image price.