Amux AI

Retrieve task

Last updated October 21, 2026

Retrieve a bytedance/seedream-5.0-pro task in the unified Amux shape, with layer info, usage and cost.

Retrieve a bytedance/seedream-5.0-pro task by ID. The response is 200 whether or not the task has finished; check status.

GEThttps://gateway.amux.ai/v1/tasks/{id}

Authorization

header
AuthorizationstringRequired

Bearer <your Amux AI key> Your Amux API key.

Request

idstringRequired

The task ID returned at submission.

Response

200response

The task's current state. output is included when succeeded.

404response

No such task, or it belongs to another workspace. Both return the same response so task IDs cannot be probed across workspaces.

Polling

Text-to-image and image-to-image usually finish within 15–60 seconds, layer decomposition in about 60–90 seconds. Start polling about 10 seconds after submission, every 5–10 seconds, or submit with Prefer: wait=60 to wait for the result.

Outputs

Image outputs are in output.images. output.videos is always present and empty; do not use it to infer the task modality.

With layer decomposition, the first item in output.images is the base image, followed by one item per layer, each with a layer field:

{
  "output": {
    "images": [
      { "index": 0, "url": "https://…/task_x-0.jpg", "layer": { "zIndex": 0, "role": "background" } },
      {
        "index": 1,
        "url": "https://…/task_x-1.png",
        "layer": {
          "zIndex": 1,
          "role": "layer",
          "bboxAbsolute": [223, 638, 685, 993],
          "bboxNormalized": [279, 448, 855, 697],
          "name": "Coffee cup",
          "description": "White ceramic cup of espresso"
        }
      }
    ],
    "videos": []
  }
}

layer uses camelCase and carries the same data as z_index and bounding_box on the Volcengine Ark native endpoint. bboxNormalized holds integers from 0 to 1000.

Output URLs point to our storage and are cleaned up after the retention period; download anything you need to keep.

Usage

This model is billed per image, so the three token fields in usage are always 0. The charge is in cost, which includes the base image and every layer for layer decomposition.

Errors

The error body matches OpenAI's; see Errors and retries for type values and retry semantics.

cURL
curl https://gateway.amux.ai/v1/tasks/{id} \
  -H "Authorization: Bearer $AMUX_API_KEY"
{
  "id": "task_01M5CWPZEAJ6FRGPDM1G7G1V29",
  "status": "queued",
  "model": "bytedance/seedream-5.0-pro",
  "created_at": "2026-10-21T08:25:44Z",
  "output": {
    "images": [
      {
        "index": 0,
        "url": "https://cdn.amux.ai/gen/20261021/task_01M5CWPZEAJ6FRGPDM1G7G1V29-0.jpg",
        "layer": {
          "zIndex": 0,
          "role": "background",
          "bboxAbsolute": [
            0
          ],
          "bboxNormalized": [
            0
          ],
          "name": "<string>",
          "description": "<string>"
        }
      }
    ],
    "videos": [
      {}
    ]
  },
  "usage": {
    "input_tokens": 0,
    "output_tokens": 0,
    "total_tokens": 0
  },
  "cost": "0.043",
  "notes": [
    "dropped seed: this model has no seed parameter"
  ],
  "error": {
    "message": "<string>",
    "code": "<string>"
  }
}