· via Hacker News – Front Page (native)
OpenAI brings back five-hour Codex and ChatGPT Work limits for Plus subscribers
OpenAI is reinstating a five-hour usage cap on Codex and ChatGPT Work for Plus subscribers after weeks of weekly-only limits, citing compute load balancing and casual users burning through weekly quotas.

Five-hour cap returns to Plus plans
OpenAI is restoring a five-hour usage limit on Codex and ChatGPT Work for ChatGPT Plus subscribers, ending a stretch in which only a weekly cap applied. According to 9to5Mac, the change was announced on X by Thibault "Tibo" Sottiaux, OpenAI's engineering lead for Codex and ChatGPT, with the limit returning on August 25.
For several weeks beforehand, OpenAI had suspended the five-hour ceiling on the two products, leaving the weekly quota as the only restraint. During that window the company also cleared users' weekly counters ahead of schedule on multiple occasions to celebrate milestones, one of which was the addition of another million active users across the now-unified platforms.
Why OpenAI is bringing it back
Sottiaux offered two justifications for reinstating the shorter cycle. He said the five-hour limit helps the company spread demand across its compute infrastructure more evenly, which in turn makes it possible to keep the weekly allowance generous. He also described the Plus audience as mostly casual users, many of them new to the tools, some of whom unintentionally burn through an entire week's allocation in one long session and are then left confused — an experience OpenAI considers poor enough to correct.
Per 9to5Mac, Sottiaux framed the move as something he had already signalled and then delayed. One group is spared: the five-hour limit will stay switched off for the coming months on the $100 and $200 Pro subscriptions. No changes were mentioned for Enterprise or Edu accounts, which run on a separate credit-based usage system.
How limits and resets work
Subscribers who hit either the five-hour or the weekly limit can wait for the next cycle to reset their usage or pay for additional credits. As 9to5Mac notes, OpenAI also resets limits for free from time to time, and users can accumulate extra resets through certain promotions and referral offers — a mechanism that became more visible during the recent run of milestone-driven weekly resets.
Why it matters
The back-and-forth is a revealing snapshot of how OpenAI balances generosity, capacity and user experience on its most popular paid tier. Lifting the short-cycle cap made the Plus plan feel roomier, but by the company's own account it let newcomers exhaust a full week of usage in a single sitting — a bad outcome for everyone, since those users then face days without access unless they pay. The reinstated limit trades that risk for a different constraint: bursty, hours-long sessions are now rationed in five-hour windows, while sustained moderate use is governed by the weekly cap.
The tier split matters too. Pro subscribers at $100 and $200 keep weekly-only limits for now, so OpenAI is differentiating plans by how usage is paced, not merely by volume. And for developers building workflows around Codex and ChatGPT Work — platforms the announcement treats as unified — the timing structure of quotas is now as consequential as their size. Expect further tuning of these cycles as OpenAI juggles finite compute against a fast-growing subscriber base.
- #openai
- #chatgpt
- #codex
- #ai
- #usage-limits