GPT-4 RPM Bugged at around 4 to 9 RPM

API
1

I have noticed that on my API account which has a GPT-4 RPM of 200 I am experiencing rate limit messages after 4 to 9 RPM on under 1,000 TPM. Other individuals have experienced the same issue. Any thoughts on what could be causing this issue? I have left a bug report message for OpenAI. Can others confirm?

My hunch is that OpenAI can’t handle the load after giving the world GPT-4 access.

3

This is a beaut. Thanks for sharing this! Will be really helpful

4

I found the issue. It was my mistake. It seems TPM is counted as prompt_tokens + max_tokens. I had assumed it was prompt_tokens + completion_tokens.

2 Likes