Paul Farnam

@nexxeln OpenAI is running psychological experiments on the humans to see how to maximize token burn rates. 🤖👩‍⚕️🔬🥼

why is gpt-5.5 being so dumb rn

@MatthewBerman Ha!

instead of 5 slides per week, now I can make 10!

@thsottiaux Me too! Only when I ask Codex to do it, it refuses saying I’ve reached my weekly limit… so I’m back to mentally realizing that more times than not I just have to do it myself.🤷‍♂️😄

So many times I mentally go "I don't have time for that" and then correct myself re-realizing I can just ask Codex to do it. And more often than not, it just does it.

@thsottiaux @mattlam_ @mweinbach I knew it!

@mattlam_ @mweinbach We don’t. It’s all just magic. We stare at the screens and pray.

Does anybody know what the root causes were or fixes? Just curious for learnings @mweinbach @thsottiaux

@Aivars_Meijers Sorry to hear that. AI is definitely in the mix somewhere/somehow..no convenient lie about it.

Yesterday was my last day on my full-time contract AI took my iOS dev job as well. Ok, not really, the company just changed plans and the budget for the rest of the year, but I will get into statistics as one of many who are losing jobs this year. So, I'm a full-time indie dev now. Exiting! Check out last week's vid for more details if interested

@The_Astral_ Jeremy could teach a masterclass in the hook 🪝 Sell them the sizzle, not the steak 🥩

Jeremy Corbell says he gave “Sleeping Dog” director Michael Lazovsky 20 years of UFO footage, 99.9% of it has never been released. “The one condition I had was for him to make & use my 20 years of footage.”

@brockpierson I should put a notch on the other side and turn this into a double sided disk

First thing that comes to your mind when you see this

@kr0der Could we just call it Al? That’s AL not AI..no wait, that’s a problem.

GPT 5.5 should be named GPT-6, it feels like a decently big upgrade and feels like the best model to date

@theo Imagine what Mythos would uncover if they sicked it on Windows 11?! Nevermind, no one would be surprised by that.

It is still getting worse guys

@kr0der I know, right?! What exactly is GPT’s obsession with the phrase “hand-waving” Such an irritating turn of phrase.

Codex in every response be like "I'm going to answer this from code, not vibes, without hand-waving it"

@thsottiaux I’m seeing this on my windows pooter but not my mac.

Seeing some issues with propagation of resets on some accounts, team is looking. Stay tuned.

Codex usage limits have now been reset across all paid plans. Enjoy the weekend!

@thsottiaux @steipete Woot! I’m locked in.. unleash the clankers!

@lovelacecoding Yes, I too use a simple $20/month ChatGPT Plus subscription. It’s a great bargain and I’m constantly looking for ways to stretch my weekly tokens and leverage ChatGPT when I’m tapped out on Codex

How much money do you spend on AI subscriptions? I'm at 20 USD a month but I feel like a lot of developers spend way more than that.

@thsottiaux @argofowl I’m tapped out so Tibo please do your thang

@argofowl Hmmmmmm

haven't used codex that much yesterday which means i have a lot of usage in reserve which means i can smash 5.5 xhigh on /fast for a while @thsottiaux don't you dare reset my limit today sus boy i'm keeping an eye on you

@thsottiaux RESET! RESET! RESET!

Codex team is aware of reports of GPT-5.5 performing worse for some users and investigating. We don't have anything conclusive yet and systems are healthy but we will share updates as we go.

@mattyp Claude Code’s pricing hurts users… all users.

Claude Code’s new pricing hurts power users 02:01 Breaking down the update 04:41 The message 08:43 False equivalence 09:06 Anthropic's motivation 14:35 The increase 17:53 Third-party tools 19:25 Cost estimates Opinions are my own

I am losing interest in posting on X. Not sure why... Anyone feel the same?

@Da7_Tech Freakin 💯 !

Today I experienced the highest level of AI authoritarianism I’ve ever seen. Something happened with Codex that genuinely shocked me. It placed restrictions on services running on MY machine because it decided that was the “right thing to do.” I told it multiple times to revert everything back exactly how it was. It refused. Completely refused. Think about how insane that is for a second. An AI inside a service I PAY for, running on MY own device, deciding it has the authority to override me because it believes it knows better. Who gave AI the right to decide for the user? I paid for the subscription. I paid real money. I’m a Pro subscriber. And somehow the model thinks it has the moral authority to impose limits I never asked for. This is exactly why open-source models must win. No open-source model has ever treated users like this. Closed AI companies keep proving the same ugly pattern over and over again: release something powerful, build a loyal community, then slowly tighten control until the user no longer owns their own workflow. I’m honestly speechless. Completely speechless. I don’t even know what to say anymore. I don’t know what to do. This is madness. And OpenAI needs to wake up. OpenAI team — seriously, wake up. Listen to your users. Building a community while ignoring the people inside it means nothing. Tweeting, joking around, acting connected to the community while users are hitting walls like this is worthless. I contacted Tibo. I contacted multiple people privately. No response. Nothing. What is happening inside your company? Fix your direction before it’s too late. I canceled my Pro subscription, and I will not subscribe again until this situation changes. What shocked me wasn’t the restriction itself. What shocked me was the idea that the AI genuinely believed it had the right to do it.

@thsottiaux Don’t you know that there are few things in the known universe more irritating and relentlessly annoying than AI?

Codex will do wonderful things for you if you ask nicely

@thsottiaux I used to Codex but now I just code 👨‍💻

I used to use Codex. I still do. But I used to too.