The moment I saw sonnet 3.7 I tried it out. It only worked once because I was metered, so I went back to 3.5, and it also displayed thinking (was it really 3.5 or a bug?)
While I was getting all output thinking, I asked the AI what it could see, and ascertained that it doesn’t even know it’s outputting bull.. bull.. .
So I couldn’t instruct the AI to cut that off, it was subconscious. Is there an API to shut it off? Because most of the time its thought process is uninteresting and takes a lot of time to spew out.