Hi, I'm trying to use stay within the tokens per minute limited enforced by the Open AI Chat Completion API.

When server responded with 429 error, I'm expecting the Retry-After field in the header exposed to the browser so the caller can retry after the server suggested period of time. The documentation here specially calls for such retry strategy

General best practices to remain within rate limits

To minimize issues related to rate limits, it's a good idea to use the following techniques:

• Implement retry logic in your application.
• ...

The problem is that the response header is set such that the Retry-After header field is hidden from the browser. See screenshot below:

This makes it difficult for the client to set the right timeout. Please consider adding Retry-After to the Access-Control-Expose-Headers field.

References:

• Understanding rate limit
• General best practices to remain within rate limits