ChatGPT bills by token count. In light of this, here is my query/queries-
- Does it mean only the user_message + system_message (prompts) token count or also the token count of agent’s reply (completion) ?
- will the system_message tokens be counted every time a query is made?
- Does the course offer any guides as to how we can keep the token count to the minimum? if not, any suggestions?
(I am not asking about per query token limit , but the overall token count over the life of a system )