"A quick letter to our friends at OpenAI"

This is supposed to be your reward for watching the first 1.5 hours 😭😭😭

oh...

watch till the end, i made a fire outro!

bro tweaking after getting 3 resets in 5 days

Theo is endless negativity.

This is one of the dumbest, most X-brain-rotted things Theo has ever said. There may be comms/onboarding/UX issues, yes, as w/ all these companies... But 'you are so off and going to lose because you have power user features that use credits more than before' is not it.

I think this guy is the most entitled sounding larp

@DiamondEyesFox I’ve found the exact opposite happening been using 5 times more usage than ever before when I used to hit limits all the time with 5.5

He is such a drama frog

They gained 3 million customers in like 3 days. Did they mess up some implementations? Sure, but this take is also objectively wrong with what we are currently seeing.

Bro just loves to bitch and yap

Isn’t codex as a whole a copy of claude code?

wrong, codex was started before claude code even existed, they just launched it later.

Deadass? Fairs you probably know that better than me lol

29h into an Ultra goal with Sol starting at the last reset and I am at 59% on 20x, I think I'd get maybe 10h on a 20x from Anthopic lol.

Theo never fails to make me laugh

Wow you even included a lag spike in the video... how diabolical. But a nice stop-frame...

@theo im now realizing you kinda look like @mradamscott

Pure waffle. What the fuck is he on about 😂 it’s felt unlimited all this week

He's right - For sure limits got nerfed. I was on the $100 plan and I could not use up my limits no matter how hard I tried. Started using 5.6 Sol extra high and had to upgrade to $200, and even there I'm blowing through my limits in... roughly 4 or 5 days.

Maybe their decisions just lack the human touch and they rely on AI too much

Theo needs to join OpenAI as Developer Advocate 😅

很遗憾这个内容已经过时了，如果你用过更新后的版本，200 刀套餐真的很难用完

I really like how @theo went from "Codex is a great harness, Claude Code is piss poor" to "I rather use GPT in Claude Code" within.... I guess it's a month?

How huge is the lamp behind theo!

exactly i fully gutted the system prompt in lee101/codex its one of the worst things you will read the openai original one is insane and waste of tokens and all sorts of annoying on purpose stuff like if you notice any changes in git stop and confirm if they are supposed to...

L

Tell this Theo guy to stop crying all the time Bro looks like an emotional m2f

GPT 5.6 is working very well in Nocturnal with the way subagents are implemented in it. It is definitely more prone to use skills and subagents than any model seen yet, but we didn't have to make any special adjustments for it to work (aside from the quirky way they're sending the reasoning summaries now).

PREACH

name drop