Thibault Sottiaux, who leads Core Products & Platform at OpenAI, wrote on X that a poll combining a “2x” proposal for Pro 100/200/500 with the return of five-hour usage limits drew less support than he had expected. The response points to a gap between the provider’s goal of smoothing out concentrated demand and users’ desire to use the service when they want.
More usage capacity is not always welcome
Sottiaux had expected to see more support for the “2x” option in exchange for bringing back five-hour limits. But the poll’s response did not match that expectation. The focus of his post is that a proposal to “increase” usage alone was not enough to win users over.
The post does not specify what exactly would be doubled or how the votes were distributed among the options. It does, however, show that an increase in total usage was presented as part of a trade-off that also involved accepting time-based restrictions. Users have to consider not only what they would gain, but also what they would give up in return.
An AI service’s ease of use is not determined solely by how much users can access it. Whether they can keep working when they need to also affects how much value they get for the price. This response offers a way to think about that distinction.
More total usage may not make up for interruptions mid-task
For example, someone finishing a presentation before a deadline or making a concentrated effort to fix code may not find usage capacity available at another time useful for the work at hand. This does not describe the circumstances of those who voted, but it could explain why time-based limits may be unpopular.
By contrast, someone who uses the service in short sessions between work or study may place more value on increased overall usage than on avoiding short-term limits. The same change can be received differently depending on how people use the service.
Even if the provider sees a change as “increasing the amount people can use,” users may see it as “restricting when they can use it.” If the terms change without bridging that perception gap, frustration over interrupted work could overshadow the benefit of increased usage.
When dissatisfaction becomes a reason to switch
When considering customer churn, the key question is not just whether limits exist, but whether users can plan their work. If they cannot predict how much they will be able to use the service, or are repeatedly stopped when they need it, they may have reason to consider an alternative.
Paid services in particular are judged on how reliably they can be used in exchange for the fee. If explanations about increased usage capacity do not match the actual experience, users may feel that the change has made the service less convenient, even if it was intended as an improvement.
What this post shows is a mismatch between the support for the proposal and an executive’s expectations. It does not provide figures on the scale of cancellations or moves to other providers, but misreading what users value is an issue that cannot be ignored when considering continued use.
Tibo’s challenge: balancing demand with user freedom
Known in tech circles as “Tibo,” Sottiaux leads OpenAI’s Core Products & Platform and is a central figure in Codex development. He is also known for having announced a plan to release some kind of improvement every day for 28 days and, if he failed, fully reset usage limits for all users, amid concerns over usage restrictions.
While signaling a commitment to continued improvements, the poll also revealed that the provider’s idea of an improvement does not necessarily match the terms users want. The issue is not only increasing usage capacity, but also how much freedom users retain to make concentrated use of the service.
Sottiauxin this postexplained that the purpose of bringing back five-hour limits was to smooth out concentrated demand. The proposal combines “2x” usage on Pro 100/200/500 with the return of time-based limits.
Comments