For the client.chat.completions.create endpoint (using the Openai python SDK), the chat completion API is giving the following error for vision based analysis while using the gpt-4o and gpt-4o-mini models.
The error response we are getting is as provided below
{
“error”: {
“message”: “Timeout while downloading {{s3_singed_url}}”,
“type”: “invalid_request_error”,
“param”: null,
“code”: “invalid_image_url”
}
}
We are getting the error response above within two seconds (meaning there is no delay in showing the error message) after the call is made.
The API is not able to access the image hosted on AWS S3 via signed URL. For the signed URL even after setting the duration for 24 hours, we still see this error.
We are seeing this happen frequently (8 out of 10 calls are failing) where, for the same url, sometimes the request is successful and sometimes it’s not.
Has anyone else faced a similar sort of issues and how have the found a workaround of this issue?
Any help in this regarding is appreciated.