This is an intentional decision in our products. Same thing goes for ChatGPT for Excel, Sheets and Powerpoint. If you run out of limits in the middle of a task, why would we want you to have a half built spreadsheet or deck?

@mattlam_ Lmao I don't think we can afford to allow people to run things for weeks at a time, goal contains a collection of individual "turns". It's not one continuous turn

@Heberth26 I literally wrote the code lmfao

@reach_vb that's very kind of you

@ryanbrewer /goal codex do anything you need to do; don't stop until you've made us $10 Million so we can pay back openAI for the tokens this is gonna use.

@ryanbrewer this is why people love OpenAI

@mattlam_ I think that's clear though right? Users don't expect unlimited credits for a /goal IMO

@ryanbrewer People were just used to Claude screwing them over w that behavior 🤣

@ryanbrewer It’s clear to me, but I’ve seen crazy comprehensions from ppl, and if they’re non devs or don’t know how /goal works, might misread it

@madmaxbr5 based

@ryanbrewer Wasn’t this removed for /goal in codex? Thought i saw it in a coded update awhile back.

Unless if you mean anything but /goal

@ryanbrewer @deyan4 Not debatable for whom? No way you guys are that clueless. Spend 5 minutes on reddit, I don't know a single person who doesn't agree that 5.3 is superior

@reach_vb @robinebers The worst feeling was using Claude and it just abruptly stop because another agent used 1 gajillion tokens

@ryanbrewer insane feature ngl

@ebenezerDN If you run out of your weekly limit it takes from the big token bank lol

@deyan4 5.5 much better at coding it's not particularly debatable either

@FreeDrThug @ryanbrewer @deyan4 Are you suggesting these posts aren't real? What is anyone gaining by saying this? What am I gaining by saying this? Lmao

@FreeDrThug @mrmarmalade11 @deyan4 type of stuff I gotta deal with every day on this ap

@ryanbrewer love it, hope everyone follows your foot steps similar to the 20$ paid plan tier :D

@mrmarmalade11 @FreeDrThug @deyan4 I'd be interested to see if you can create a benchmark that can show this