henry 发自 凹非寺
Quantum Bit | Public Account QbitAI
In my opinion, OpenAI (SI) has gone mad! They just act up during holidays～
Just now, Cyber stepfather and OpenAI Codex head Tibo said: In the next 28 days, you need to choose one option every day.
Either deliver a improvement that is clearly useful to most Codex/Work users, or perform a full quota reset.
In all fairness, the quota reset is easy to understand, but the word “improvement” is really clever. It could be a new release, or it could just be fixing bugs.
However, regardless of what is served each day, at least according to this statement, we have a minimum guarantee:
Experience if there’s an improvement; reset if there isn’t. Either way, you’re not losing anything.
And if you watch it a lot, you’ll find that this choice between two options has become Tibo’s standardized playing style.
Honestly speaking, for users, whether it’s new features or quota resets, it’s always a good thing.
Just resetting so frequently—first, the quota may not be used up; second, it also inevitably causes people to wonder:
Is OpenAI’s eagerness to please users really a sign of strong service awareness, or is it because they don’t feel so confident in their own product?
Still... Is the pressure from Claude Opus 5.5 too much?
Tibo online collects user feedback
The background of this matter is that Tibo has been soliciting feedback on product improvements from netizens recently:
What’s missing in the Codex? What is something that everyone has wanted but hasn’t used yet?
After collecting a circle, he locked down the next work in four directions:
Simplify products, improve efficiency to support more usage, introduce advanced features, and launch new models.
However, just as Tibo was indulging in his community art, Lauren Tan from the adjacent Grok Bot came over to interact with him.
She first mocked OpenAI for producing “wasteful” content, relying entirely on resetting limits to keep up the act.
It’s also said that OpenAI’s promotion is extremely strong, and its quality is outstanding.
Tibo naturally didn't hold back either; his reply was basically: You can do it, go ahead～
The exchange between the two parties has a bit of a mutual draining effect. But beyond the excitement, users’ dissatisfaction with the OpenAI product experience is not just about “not having enough quota”.
Let’s organize it, focusing mainly on three aspects:
First of all, DevDay presented over 20 updates at once, so much that it’s hard to figure out what’s the main focus. The only key point is that the personal Agent Dot currently seems to still be affected.
Secondly, resetting is intended to serve the work usage — so that people don’t feel any lag when actually working.
But this frequent, unpredictable reshaping makes it difficult to plan specific work schedules in advance.
Not to mention, the quota packages are still decreasing, and the new $500 package is too expensive~
Finally, it’s also the model issue. The connection was cut mid-task, and no matter how many times you reset it, it still gets stuck.
It’s like you have an infinite supply of bullets, but each time, they get stuck and can’t be fired. In other words, having an infinite supply of bullets is completely useless.
So, in Tibo’s promise this time, what’s most important probably isn’t “every day”, but “clearly useful to most users”.
One less disconnection, one less round of rework, one less research on an entrance – all that counts as improvement. There’s no need to come up with another big trick.
After 28 days, it’s best for the stepfather to minimize people’s attention to his activities.
Reference link
[1] https://x.com/thsottiaux/status/2106845241357824205