Using Webhook Endpoints for Async Results

Webhook endpoints can be used to obtain asynchronous results from your requests. This is useful when your model takes longer than the max timeout (180 seconds) or when you don’t want to wait for the model to finish processing the request and prefer receiving the response once the processing is completed.

To use this feature, simply include the webhook_endpoint field in your request body, assigning it to your desired URL. As soon as this is done, your requests will return immediately with the run_id. After processing, a POST request containing the results and run_id will be made to the provided webhook_endpoint URL.

​

Example curl request

Here is an example curl request with the webhook_endpoint field included:

curl --location 'https://run.cerebrium.ai/dreambooth-webhook/predict' \
--header 'Authorization: <YOUR_PUBLIC_API_KEY>' \
--header 'Content-Type: application/json' \
--data '{
    "prompt": "An astronaut riding a horse in space",
    "webhook_endpoint": "https://your-webhook-url.com"
}'

​

What to expect

When your model has finished processing, it will send a POST request to https://your-webhook-url.com with the results and the run_id.

The response returned from your initial request will look something like this:

{
  "run_id": "dummy-run-id-9ac612631200",
  "message": "We will return the images to the endpoint provided when the task is complete."
}

The result sent to the webhook will look something like this:

{
  "run_id": "dummy-run-id-9ac612631200",
  "run_time_ms": 5980.915546417236,
  "message": "Successfully generated images",
  "result": ["..."]
}

To test webhook endpoints, https://webhook.site/ provides a great platform. You can create a free URL to receive your POST requests and inspect the results.