Errors
Ce contenu n’est pas encore disponible dans votre langue.
Client API failures use the error schema documented in the API Reference. Always branch on the HTTP status and the machine-readable code; keep the human-readable message for diagnostics only.
{ "success": false, "message": "Assistant not found", "requestId": "00000000-0000-4000-8000-000000000000", "errors": [{ "code": "ASSISTANT_NOT_FOUND", "message": "Assistant not found" }]}Typical status classes are:
| Status | Meaning | Client action |
|---|---|---|
400 |
Invalid input | Correct the request before retrying. |
401 |
Missing or invalid credential | Check the service-specific API key. |
404 |
Resource does not exist | Verify the ID and account context. |
409 |
Current state prevents the operation | Read the error code and resolve the conflict. |
429 |
Client API request limit reached | Wait for Retry-After, then retry. |
5xx |
Service-side error | Retry with bounded exponential backoff. |
The Public RAG service has its own, simpler error format. Use only statuses declared by that service’s operation in the Reference.

