OpenAI has reinstated its five-hour usage limits for Codex and ChatGPT Work on the Plus plan, ending a stretch of several weeks in which Plus subscribers were governed only by a weekly usage cap.
According to 9to5Mac, whose report by Marcus Mendes was picked up by Techmeme, the company announced the change and said the tighter cadence would help smooth out the load on its compute. NDTV Profit and Ukrainian outlet Mezha carried the same news, and Notebookcheck characterized the move as an abrupt return to "harsh" limits.
The practical difference comes down to pacing. A weekly cap lets a subscriber burn through a large chunk of their allowance in a single marathon session — say, a long day of refactoring code with Codex — and only feel the ceiling later in the week. A rolling five-hour window forces usage to be spread out: hit the limit and you wait for the window to reset, regardless of how much of your weekly budget remains untouched. For OpenAI, that spreading-out is the point, since demand spikes are what strain the finite pool of chips serving every user at once.
None of the sources here specify the exact numeric allowance inside the five-hour window, or whether the weekly cap remains in place alongside it. What is clear is the direction of travel: OpenAI removed a constraint, watched what happened, and put it back.
Why it matters: coding assistants are becoming load-bearing tools in professional workflows, and this is a reminder that the terms of access can tighten with little notice — the compute behind them is a scarce, expensive resource, and the people paying for a subscription do not fully control how much of it they get.