Marigold V2 API: Pricing, Documentation

Authentication

All requests require an API key passed via header.

Webhook (Optional)

Instead of polling, you can receive a Webhook callback when the request reaches a terminal state. Provide a Webhook URL via header on the submit request.

Example: enable Webhook

curl -X POST 'https://gateway.pixazo.ai/marigold-v2/v1/predict' \
  -H 'Content-Type: application/json' \
  -H 'Ocp-Apim-Subscription-Key: YOUR_SUBSCRIPTION_KEY' \
  -H 'X-Webhook-URL: https://your-server.com/webhook' \
  --data-raw '{"image_url": "https://storage.googleapis.com/falserverless/example_inputs/dog.png", "prediction_type": "depth"}'

Callback Payload (success)

{
  "request_id": "marigold-v2_019dxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "status": "COMPLETED",
  "model_id": "marigold-v2",
  "output": {
    "media_url": [
      "https://pub-582b7213209642b9b995c96c95a30381.r2.dev/v1/{request_id}/output.png"
    ],
    "media_type": "image/png"
  },
  "created_at": "2026-09-22T11:56:21.804Z",
  "completed_at": "2026-09-22T12:01:44.210Z"
}

Failure callback shape

{
  "request_id": "marigold-v2_019dxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "status": "ERROR",
  "model_id": "marigold-v2",
  "error": "Description of the failure"
}

Delivery semantics

• terminal mode: one Webhook callback when the request is COMPLETED or ERROR.
• sync mode: a Webhook callback on each status change.
• Callbacks are idempotent on request_id — de-duplicate on it.
• Respond 200 within a few seconds; the Webhook endpoint must be HTTPS.

Request Parameters

Good to know

• Exactly one map is returned per request. There is no batch option: num_images, n, count, batch_size and number_of_images are rejected.
• To get several maps of the same image (for example depth and normals), send one request per prediction_type.
• The output is a PNG visualization at the processing size — good for viewing, previews and as an input to other image tools, but not a metric (absolute-distance) depth measurement.
• The cost is the same for every prediction_type and image_size; the current rate is in the pricing panel on this page.

Request Headers

Response Handling

Only what can be judged from the request body itself is rejected synchronously. Everything that needs an input to be fetched or the model to run is reported through the status endpoint instead, as status: "ERROR". Failed requests are not billed — the hold is released in full.

Retrieving Results

Poll the status endpoint with the request_id from the submit response until status is COMPLETED (or FAILED/ERROR), then download the model from output.media_url. Processing usually takes from a few seconds to a few minutes, so poll every 10–15 seconds rather than continuously.

curl 'https://gateway.pixazo.ai/v2/requests/status/marigold-v2_019dxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx' \
  -H 'Ocp-Apim-Subscription-Key: YOUR_SUBSCRIPTION_KEY'

Completed response

{
  "request_id": "marigold-v2_019dxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "status": "COMPLETED",
  "model_id": "marigold-v2",
  "output": {
    "media_url": [
      "https://pub-582b7213209642b9b995c96c95a30381.r2.dev/v1/{request_id}/output.png"
    ],
    "media_type": "image/png"
  },
  "created_at": "2026-09-22T11:56:21.804Z",
  "completed_at": "2026-09-22T12:01:44.210Z"
}

The result is a PNG visualization of the requested map (for depth, a colorized depth map). It is meant for viewing and as an image input to other tools; it is not a lossless numeric depth tensor.

Response Fields

Status Values & Flow

QUEUED → PROCESSING → COMPLETED (success) or FAILED/ERROR (failure).

How billing is measured

Billing is per generated map, and every request returns exactly one, so the amount held when the request is submitted is the amount charged on completion. The price is the same for every prediction_type and image_size. If the request fails, the hold is released in full.