If you use OpenAI Codex enough, you eventually start thinking about usage limits. You might be in the middle of fixing a bug, adding a feature, or letting Codex tear through a larger project when you realize your precious allowance is disappearing faster than you would like. Suddenly, you start wondering whether that next task is really worth it.
Then Tibo appears with another reset, and everything is right with the world again.
If you spend any time around the Codex community, you probably already know who I am talking about. Tibo is Thibault Sottiaux, an OpenAI product leader who has somehow become something of a rockstar among Codex users. On X, where he goes by @thsottiaux, he has amassed more than 500,000 followers.
Calling him simply an OpenAI employee doesn’t really do him justice. OpenAI previously identified Sottiaux as Head of Codex and described him as a Member of Technical Staff who led the company’s coding agent. His responsibilities have since expanded, with OpenAI more recently identifying him as Core Products Lead and Head of Product & Platform.
In other words, Tibo isn’t some random OpenAI employee who stumbled across a big red reset button. He has been one of the people actually leading Codex and now appears to have much broader responsibilities within the company.
And yet, despite all of those impressive titles, I mostly know him as the guy who gives me more Codex.
So, what’s with all the resets?
OpenAI puts usage limits on Codex depending on your ChatGPT plan and how heavily you use the service. That makes sense. AI models cost money to run, and OpenAI can’t realistically let everyone hammer away at them forever.
That doesn’t make hitting a limit any less annoying, though. When Codex has become part of how you actually develop software, watching your remaining allowance shrink can influence what you ask it to do.
This is where Tibo comes in. Every so often, he shows up bearing gifts in the form of a usage reset.
Sometimes there is a problem with usage or metering. Other times OpenAI is celebrating a Codex milestone. Whatever the reason, the outcome is what users care about: the meter gets replenished, and everybody can get back to work.
It has happened often enough that Tibo resets have basically become part of Codex culture. And look, Tibo knows exactly what he has created, too. He once asked users why they had switched to Codex and what OpenAI could improve, while specifically telling them not to say “reset.” Yeah, good luck with that.
I actually plan my Codex usage around Tibo
I am not observing this phenomenon from a distance. I have notifications turned on for Tibo’s posts on X, and while I certainly want to know about Codex news, I would be lying if I said potential resets weren’t a big reason why.
I have even started planning my Codex usage around his reset promises. If Tibo says another reset is coming, I know I can loosen the reins a bit and let Codex burn through more of my remaining allowance. If I am getting close to the limit and Tibo hasn’t promised anything, I become more conservative.
Yes, I realize how ridiculous that sounds. I am a grown man planning my AI usage around the social media posts of a guy at OpenAI. But hey, it works.
Codex has become a tool I use extensively for software development. I don’t just occasionally ask it for a snippet of code. I use it to work through actual projects, diagnose problems, implement changes, run tests, and handle the sort of tedious development work that can consume a lot of usage.
When you use Codex that way, the limits stop being an abstract number on an OpenAI support page. You notice them.
Tibo may be keeping me from paying for ChatGPT Pro
There is another funny part to all of this. Tibo may actually be costing OpenAI money, at least when it comes to me.
I currently subscribe to ChatGPT Plus. I use Codex heavily enough that I have considered moving to ChatGPT Pro, particularly when I am working on a project and see my remaining allowance getting uncomfortably low. For the most part, though, Tibo has helped keep me on Plus.
Just when I start thinking that maybe I finally need to spend the extra money on Pro, another reset arrives or Tibo promises one is coming. Suddenly I have enough Codex capacity to keep working, and that expensive upgrade doesn’t seem quite as necessary anymore. Thanks, Tibo. My wallet appreciates you.
There is a legitimate business question buried in this silliness. Heavy Codex users seem like exactly the people OpenAI would want to convert from Plus to Pro, yet the company keeps handing out additional capacity that can delay that decision.
Maybe OpenAI believes keeping developers building with Codex is more important right now than squeezing additional subscription revenue out of them. Maybe the resets are simply a way to celebrate growth, compensate for problems, and keep the community happy. Whatever OpenAI’s reasoning, I know the resets have increased the value of Plus for me.
Somehow, Tibo became an OpenAI rockstar
This is the part I find genuinely surreal. OpenAI is one of the most closely watched technology companies on the planet. It has famous executives, prominent researchers, and products used by hundreds of millions of people.
Yet here is an OpenAI product leader with more than half a million X followers becoming a minor celebrity because developers genuinely look forward to hearing from him. That isn’t normal, and I mean that in a good way.
You expect technology rockstars to be CEOs, founders, YouTubers, or developers who have spent decades building a public persona. Tibo doesn’t fit that mold. His X bio is remarkably simple: “Codex & ChatGPT @OpenAI.” Apparently, that’s enough.
His background makes the situation even more interesting. Before joining OpenAI, Sottiaux worked at Google DeepMind, where OpenAI says he led human data work for Gemini and built AI and machine learning workflow infrastructure for DeepMind research. Earlier in his career, he worked as a software engineer at Google and held applied research roles.
This isn’t someone OpenAI hired to become an influencer. He is a deeply technical product leader who somehow became one anyway. And I think I understand why.
Huge technology companies can feel faceless. You pay your subscription, use the product, complain into a support form when something goes wrong, and hope somebody somewhere eventually notices.
Tibo feels different. He talks directly to Codex users, asks what they are doing with the product, acknowledges problems, shares what the team is working on, and occasionally shows up like Santa Claus with another sack full of usage.
That creates a surprisingly personal connection between the people building Codex and the people using it.
OpenAI has previously said Tibo uses Codex heavily himself, including for more than 100 tasks per day. That probably helps explain why his posts often feel closer to how serious Codex users actually work than the polished product messaging you normally get from a huge technology company. He’s not just talking about the product. He’s using the damn thing.
Long may Tibo press the reset button
I don’t expect the reset party to last forever. AI inference costs real money, Codex usage isn’t free for OpenAI, and there will presumably come a day when the company decides we have all been spoiled enough.
Until then, I am keeping those X notifications turned on. If my Codex allowance starts getting dangerously low and Tibo says a reset is coming tomorrow, you can bet I am going to plan accordingly.
So, who is Tibo? His name is Thibault Sottiaux, and he is an important OpenAI product leader who helped lead Codex before taking on broader responsibilities across the company’s products and platform.
But that doesn’t fully explain who he has become to the Codex community. Somehow, this deeply technical OpenAI employee has become a rockstar, a patron saint of depleted usage meters, and the man hundreds of thousands of developers are happy to see pop up in their notifications.
To me, though, he’ll always be the guy with the reset button. Long may he press it.
Support independent tech journalism
NERDS.xyz is independently owned and operated. If you enjoy my coverage of Linux, AI, hardware, cybersecurity, and tech culture, consider supporting the site on Ko-fi.
Support NERDS.xyz