Skip to main content
When a call fails, save the Request ID and the complete error text before changing settings or retrying.

​
Setup and configuration

  • Where should I start? Read the quick start.
  • Which Base URL should I use? It depends on the client; see the integration overview.
  • Why do I see 404? Check for /v1/v1/... and whether the client appends a path automatically.
  • Why does a new API key fail? Check its status, credit, token group, model ID, and Authorization: Bearer format.

​
Models and charges

  • How do I know whether a group can use a model? Filter the Model Marketplace by the group and copy the exact ID.
  • How much did one request cost? Use the console log and read usage and billing.
  • Can an error be charged? It can be, if an upstream request already ran or the client sent a new request. Reconcile by Request ID.
  • Why is the balance normal but the key says insufficient credit? Check account credit, key limit, expiry, and group.

​
API errors

  • What do 400, 401, 429, 500, 502, 503, 504, or 524 mean? See API error guide.
  • Why can a long conversation no longer continue? Save the error, start a new session, and avoid repeatedly submitting the same request.
  • Why is image generation slow or timed out? It is synchronous. Keep the Request ID and do not immediately submit the same job again.
  • Why did an error later appear to succeed? Do not assume the platform retried it; check client retries and matching Request IDs.

​
Get help efficiently

Use the self-check guide, then provide the Request ID, time and time zone, client and version, model, endpoint, full error text, and whether the issue is intermittent or persistent. Never send a full API key, password, or private content.