I see the same behavior, a chat completion API call using gpt-3.5-turbo with 2k request tokens takes on average 150s, which is longer than the ChatGPT Plus web app. This is not network-bound.
1 Like
I see the same behavior, a chat completion API call using gpt-3.5-turbo with 2k request tokens takes on average 150s, which is longer than the ChatGPT Plus web app. This is not network-bound.