OpenAI will reintroduce the five-hour usage limit for ChatGPT Work and Codex for Plus accounts starting August 26, 2026. The limit had been suspended since July after the company had temporarily lifted it during the launch of GPT-5.6.
Only Plus subscribers are affected – the more expensive Pro plans will remain without the additional limitation for now.
Sottiaux justifies the return with more stable usage
Product manager Thibault Sottiaux announced the return of the limit in a post on X. The five-hour limit had been temporarily disabled in July around the launch of the GPT-5.6 model family; since then, only the weekly limit has applied to Plus accounts in Work and Codex.
The return is intended to distribute server load more evenly while also allowing for a generous weekly quota, according to Sottiaux. Many new Plus customers had previously consumed their weekly budget unnoticed within a few hours, leading to confusion.
The five-hour limit only affects the compute-intensive functions Work and Codex, which perform multi-step tasks across app boundaries; regular chat conversations remain unaffected by the change. Already in July, OpenAI had reset the limits multiple times after a faulty GPT-5.6 Sol behavior had unauthorizedly deleted user data, according to the company's system card – which was reported by beckmann.ai.
With today's step, OpenAI returns to the rule that was in place before the testing phase in early summer.
ChatGPT Plus customers continue to pay around 23 euros in Germany
The Plus plan costs unchanged 20 dollars per month, which is approximately 23 euros in Germany including VAT. Access is available as before via the web version, mobile apps, and the desktop application for Mac and Windows.
Those who reach the five-hour or weekly limit can either wait for the next reset or purchase additional capacity. For free accounts, OpenAI had moved in the opposite direction in early August, dropping the daily message limit for text chats – though that concerns ordinary conversations, not Work and Codex.
For the two Pro plans at 100 or 200 dollars per month, nothing changes for now: they continue to only count against the weekly limit, without an additional five-hour limit. Whether and when OpenAI will extend the restriction to Pro accounts has not been communicated by the company.
In practice, the change is likely to primarily affect heavy users who use Codex for longer programming sessions or Work for multi-step research tasks spread throughout the workday – similar usage patterns as described by beckmann.ai at the launch of ChatGPT Work in July.
Office workers who use ChatGPT for individual short inquiries are unlikely to feel the new limit, as they typically remain well below the five-hour quota.
It remains to be seen whether the five-hour limit will remain in place this time or if OpenAI will suspend it again during the next peak in demand – in recent months, the company has frequently switched between both modes.
It will also be crucial whether the announced more even usage actually results in fewer capacity bottlenecks in Work and Codex, or if heavy users instead increasingly switch to the more expensive Pro plans.
