AI Commentator Reports Consuming 25 Billion Tokens In One Week

this is because of loops btw

@MatthewBerman you can’t just be pregnant forever.

at some point there needs to be a baby.

@MatthewBerman Top models used?

@MatthewBerman thats a token vacuum loop matthew

@MatthewBerman If you are going off the codex profile then that's incorrect. If you do /goal and then look at that number it's usually far far far less than what the profile reports for some reason. It's totally broken

@BradleyYoungjr yes and yes and a video

@MatthewBerman @grok how much money in api costs is that for each ai model : open ai gpt 5.5 , Claude opus 4.8 , and Claude fable 5

@McCalebRoyer GPT mostly these last couple weeks

@MatthewBerman and the quality of work?

I burned $10k in Fable and GPT 5.5 found several mistakes.

@MatthewBerman Damn bro... save some for the rest of us! ...and half of that is without Fable!

@MatthewBerman Using many tokens is easy, the key ability to develop is to use many while getting the most out of them.

@MatthewBerman on what?

@AdamYoungblood_ subs + generous providers to test their products

@RobertDMellish 🤣

@MatthewBerman Someone needs to develop a token/value metric

@floyddepalma coming soon.

@SethGammon "make every single page/modal in my app load under 50ms"

@BenSchneiderX Building something cool, will open source it soon!

@MatthewBerman rookie numbers

@MatthewBerman