Thibaultmol Currently you can only see how many tokens you got back from the api's reply but not actually how tokens were sent (including the context window and such) .
Luis It should include all tokens used in the answer "generation" step: input (user + system prompts) and response. We'll check
