Error codes
Written by
Updated at July 16, 2024
In response to any request, Foundation Models models return a message containing a request processing code and status. If there is an error processing a request, the message will also include its possible cause.
When processing model responses, in your message, refer to the request code and status. Error messages contain useful information, but may change after model updates.
| gRPC | REST | Status | Reason | Solution |
|---|---|---|---|---|
| 0 | 200 | OK |
Operation completed | — |
| 3 | 400 | INVALID_ARGUMENT |
Invalid request | Check the request format. You may have specified an incorrect model URI, exceeded the prompt length, set an invalid parameter value, or failed to comply with YandexART ethical guidelines. |
| 4 | 504 | DEADLINE_EXCEEDED |
Request timed out | Network issues between the client and server. Try again or contact support. |
| 7 | 403 | PERMISSION_DENIED |
Not enough permissions | Make sure that the service account you are using to send your request has the required roles. If access to the model you selected is provided upon request, make sure you have obtained it. When accessing a model fine-tuned in DataSphere, add the account to the member list of the project that stores the fine-tuned model. |
| 8 | 429 | RESOURCE_EXHAUSTED |
Quota exceeded | Depending on your quota, wait or contact support to increase your quota. When requesting a quota increase, specify the model, working mode, and the name of the quota to increase. Our support team may request additional information. |
| 12 | 501 | UNIMPLEMENTED |
Service-side problem | Contact support. |
| 13 | 500 | INTERNAL |
Internal service error | Contact support. |
| 14 | 503 | UNAVAILABLE |
Service unavailable | Try again or contact support. |
| 16 | 401 | UNAUTHENTICATED |
Authorization error | Check that you are using the correct account, API key, and IAM token, and that the token has not expired. |