Skip to main content

Modify the system prompt

Feature Request

Recently, some of the models (mainly ChatGPT) have started outputting results which unnecessarily waste tokens. For example, you can ask it for a simple function and, it will but, then it will also spit out a “better” suggestion and, quite likely a third “best” solution…

I now typically have to suffix all messages with:

No TokenMaxxing BS. If you are going to output a solution, rather than write it out and then suggest a better one, just output the final best solution you compute

I have added the above to my settings but, think it should be applied globally. It’s just so needlessly wasteful

Status: Completed2 comments

Log in to comment and vote

Comments2

  • Coffee Moss

    I think this is already in there. In “Anything else T3 Chat should know about you?” paste the system prompt you want in there. And you can change personalities by changing the profile.