* Some people suggest not using max reasoning due to overthinking and looping issues, this may consume more tokens than needed.