I made an financial assistant and gave it a dummy file in the file-search tool in order to answer from it, it worked normally but I wanted to test what files would the assistant use if I were to insert a unrelated file, and suddenly my token usage was more than normal. Is their a way to check the files that he used because I wanna know why was the token usage higher after I assigned a unrelated file
If am not wrong, OpenAI does not provide a direct way to audit or view which specific parts of the files were used in generating a response. The token usage reported by the API is cumulative and does not break down by file or specific content within a file.