Skip to main content

Short answer

When troubleshooting an API call, record the model name, endpoint, and call time first. Then check the HTTP response headers for x-request-id or request-id. If the endpoint returns an error response, also inspect the JSON body for identifiers such as request_id, and use the field that can be found in the APIYI logs. Wan and HappyHorse video endpoints also return request_id in the response body. The task_id in the same response is used to query the video task and is not the same as the Request ID. The id or task_id returned by asynchronous endpoints such as Seedance and Veo is also a video task ID.
The APIYI log page can search the identifiers shown in the “Request ID / Upstream Request ID / Completion ID” filter. Open the log details in the console and search using the identifier provided by the user.

Start with these three steps

1

Step 1: Confirm the model and endpoint

Record the full model name and the actual URL. Text models usually use /v1/chat/completions, embedding models use /v1/embeddings, image models may use /v1/images/generations, and video models may use /v1/videos or a model-specific asynchronous endpoint.
2

Step 2: Record the call time

Record when the request was sent and include the timezone, for example 2026-08-25 14:32 (UTC+8). If the request was retried, record the approximate time of each retry.
3

Step 3: Save the response and log details

Save the HTTP status code, complete response headers, complete response body, and the client exception. Then open the APIYI “Logs” page and search by Request ID, Upstream Request ID, or Completion ID.

Where to find the Request ID by API type

HTTP header names are case-insensitive, but the separators in the field names matter: x-request-id, request-id, and request_id are different spellings. Check both the response headers and error bodies instead of searching for only one field name.

Read it in code

Python

cURL

Use -i to print both response headers and the response body. Look for x-request-id or request-id in the headers:
For Wan or HappyHorse video calls, also save the two body fields:

Search the console logs

1

Open the call logs

Sign in to the APIYI console, open “Logs”, and open the details of the call you want to investigate.
2

Enter an available identifier

Paste the identifier into the “Request ID / Upstream Request ID / Completion ID” filter. Prefer the Request ID from the APIYI response headers; if it is unavailable, try the Upstream Request ID or Completion ID.
3

Compare the details

Compare the model, call time, endpoint, channel, HTTP status, error code, and billing record to determine whether the request reached APIYI, reached the upstream provider, and should be retried after changing the request.
If a user only knows the approximate failure time, you can still narrow the search by model, endpoint, and time. A Request ID usually identifies a single call much faster.

Why might there be no Request ID?

If the request fails before an HTTP response is received—for example, because of DNS resolution, TCP/TLS connection setup, a local proxy rejection, or a client-side connection timeout—APIYI has no opportunity to return response headers. In that case, there is no Request ID available to the client. Provide the following instead:
  • The original client exception and complete stack trace;
  • The request time and timezone;
  • The model and endpoint;
  • The HTTP client, proxy, or network environment;
  • A Request ID from a successful or retried call, if available.
Do not send a complete API key in troubleshooting material. Mask the middle of the key, and do not expose private data or a request body containing full business content or images.

What to provide to support

  • Request ID;
  • Upstream Request ID or Completion ID, if shown in the logs;
  • Model name and endpoint;
  • Call time with timezone;
  • HTTP status, complete response body, and client exception;
  • A redacted request and the error code and billing status shown in the console logs.

How Do I Troubleshoot Model Errors?

Troubleshoot parameters, groups, timeouts, and model errors

How Can I View My Call Logs?

View API calls, errors, and billing details in the console

Log Query API

Query call logs by time, model, or request_id

Image API Best Practices

Troubleshoot image timeouts, disconnects, billing, and request IDs