Troubleshoot a failed API request

Check your account, endpoint and error message before changing your configuration. Updated September 27, 2026.

Start with three checks

  1. Confirm that your account has allowance and the plan has not expired.
  2. Check the client base URL and avoid adding /v1 twice.
  3. Send one short, non-sensitive request and note the error status.

400 · Request format

Check the required fields and supported parameters. Reduce the request to a short text-only example before adding optional features.

401 · Authentication

Check that the key belongs to this service, has not been revoked, and is configured for the correct endpoint. Remove conflicting authentication settings without printing the key.

403 · Permission

Check account and key permissions. Repeating the same request will not resolve a permission restriction.

404 · Path or identifier

Check the base URL, request path and the identifier supported by your account. A missing /v1 or a duplicated /v1 can cause a wrong path.

413 · Request size

Reduce attached content and conversation context, then retry a smaller request.

429 · Allowance or rate limit

Read the returned error message. Check allowance and expiry, reduce concurrency and wait before retrying. A rate limit and an exhausted allowance need different fixes.

500 / 502 / 503 / 529 · Service failure

The service or upstream may be unavailable. Wait and retry a short request with a limited retry count. If the error persists, send its time and status to support.

Next steps