Hey, I am using gpt-4o vision + streaming, and I want to track how much tokens my users have used for analytics. When user closes the stream beforehand, I don’t have a last message with the tokens_used parameter. Is there any way to assign a request_id to the api request, and later query the tokens_used for the specific request?
I guess it is not a question, since I know the answer is no. It is more of suggestion. I believe the ability to query the token_usage would simplify life for many OpenAI users, at least that is my impression after reading this forum for some time.