But let’s calculate the value instead: e.g. a CI / CD pipeline with sharded builds including installing 3 runners with docker cache, firewall, etc. = 1 out of 200 tasks per day.
That CICD pipeline done by a developer without AI would have taken at least 30 days with a daily invoice of ~1000$ minimum.
That means 200 issues - all worth ~150k on average = 30 million $ saved … per day!
And the quality of the output is also higher than what human coders would have done. Except for frontend development. AI models just suck in that. And what I saw from Fable 5.1 or the newest gemini flash or Astra that won’t change anytime soon. It looks to me like they are just faking it when there is good stuff coming out.
So if you are looking for a career in IT: become a frontend engineer! AI won’t replace that!
So a guy who knows how to use codex can save you hundrets of millions per month if you have that need!
i still cannot get past this i cannot get any work done at all. just to set up a project takes longer than this.
You can do that the legal way - it is called a custom gpt.
You can configure a server and then point your CustomGPT to it’s openapi definetion.
Before ChatGPT hat CustomGPTs they had plugins (I had my first autocoder plugin back then) - I think that was in the middle of 2023.. then they dropped plugins and replaced it with CustomGPT .. and later introduced plugins again - but they are someting else now.
huh? how is this answer over your last comment 
im going to write a bridge for this, please dont ban me openai, please give me a job instead if you are reading this.
It is very simple, bridge the normal chat to codex.
Sorry the reset didn't restore the allowance as expected. Please refresh Settings → Usage and confirm what usage is shown there.
Also, thanks @jochenschultz for the helpful input and for pointing out the session ID details.
If the allowance still looks incorrect, please contact Support and include the approximate time you applied the reset, your session ID, and what the Usage page showed immediately afterward. That should help the team investigate what happened.
You can also review the current guidance on banked resets here: https://help.openai.com/en/articles/20001498
— Farah
