OpenAI has reinstated a five-hour usage limit for ChatGPT Plus subscribers using ChatGPT Work and Codex, reversing a temporary relaxation that had allowed eligible paid users to rely only on their broader weekly allowance. The change takes effect August 25, 2026, and means Plus users can once again encounter a short-term usage cap before their weekly allocation is exhausted.
The five-hour restriction had been temporarily removed in July for paid users of ChatGPT Work and Codex. OpenAI engineering lead Thibault “Tibo” Sottiaux said the company decided to restore the limit to smooth computing demand and preserve generous weekly usage while also preventing relatively casual and newer Plus users from unintentionally consuming most or all of their weekly allowance during a concentrated period of use.
OpenAI Brings Back The Five-Hour Usage Cap
The reinstated restriction specifically affects ChatGPT Work and Codex usage on the Plus plan. It should not be interpreted as a new five-hour limit on every ChatGPT conversation or every feature available to Plus subscribers.
The change follows several weeks during which OpenAI had paused the shorter usage window. During that period, users could continue using the relevant tools without hitting the five-hour interval cap, although a separate weekly usage allowance remained in place.
The five-hour window effectively creates another layer of usage management between individual sessions and the overall weekly allowance.
What Has Changed?
| Usage Policy | Temporary Arrangement | From August 25, 2026 |
|---|---|---|
| ChatGPT Work — Plus | 5-hour limit removed | 5-hour limit restored |
| Codex — Plus | 5-hour limit removed | 5-hour limit restored |
| Weekly allowance | Remained | Remains |
| Pro $100 | 5-hour limit temporarily disabled | Remains disabled for now |
| Pro $200 | 5-hour limit temporarily disabled | Remains disabled for now |
| Enterprise/Edu | Separate usage system | No change announced |
| Additional credits | Available where applicable | Remain an option after limits are reached |
OpenAI has not indicated that the weekly allowance itself is being removed or replaced. Instead, the five-hour restriction is being reinstated as an additional mechanism for managing access.
Why Did OpenAI Restore The Limit?
OpenAI’s explanation centers on computing capacity and user experience.
Sottiaux said the five-hour limit helps smooth demand on OpenAI’s computing infrastructure. Rather than allowing users to consume a large amount of their available capacity in a short period, the shorter window spreads usage over time.
This becomes increasingly important as AI products move beyond conventional question-and-answer interactions toward agentic workloads. Codex can perform coding tasks, while ChatGPT Work is designed for longer-running research and deliverable-oriented work. Such workloads can require substantially more computing resources than a short conversational exchange.
OpenAI’s Two Main Reasons
| Reason | What It Means |
|---|---|
| Compute management | Usage can be distributed more evenly instead of creating sharp demand peaks |
| Weekly allowance preservation | OpenAI can maintain a larger overall weekly allocation while controlling short-term consumption |
| User experience | Casual users are less likely to accidentally consume a large portion of their weekly allowance |
| Capacity planning | Shorter usage windows give OpenAI greater control over infrastructure demand |
The company therefore presents the five-hour limit less as a reduction in the overall generosity of the Plus plan and more as a mechanism for controlling how quickly users can consume available capacity.
What It Means For ChatGPT Plus Users
For most casual Plus subscribers, the impact may be limited. Someone who uses Codex or ChatGPT Work occasionally may never reach the five-hour threshold.
The change becomes more significant for developers, researchers and other users who run intensive tasks for extended periods. A user conducting several coding or research sessions in a concentrated period may encounter the five-hour restriction even when they still have capacity remaining under the weekly allowance.
This distinction is important because the five-hour limit and weekly limit operate on different timescales.
Five-Hour Limit vs. Weekly Limit
Weekly allowance:
Controls the total amount of eligible usage available over the broader weekly period.
Five-hour window:
Controls how much usage can be concentrated within a shorter period.
In practical terms:
Short-Term Usage → 5-Hour Limit → Weekly Usage → Weekly Reset
This means a Plus subscriber could potentially have unused weekly capacity but still need to wait for the shorter usage window to reset.
Codex Users Are Likely To Notice The Change Most
Codex is particularly relevant because it is designed for software-development workflows rather than simple conversational queries. OpenAI describes Codex as its coding agent, and the service is available across ChatGPT plans, subject to plan-specific usage limits.
Coding agents can consume resources through longer-running tasks, repeated tool calls and multi-step workflows. That makes short-term capacity management more important than it would be for conventional text conversations.
Developers who use Codex intensively may therefore need to plan workloads around the five-hour window rather than simply working until their weekly allowance is exhausted.
For users who rely on Codex for professional work, the change could encourage more deliberate task scheduling, particularly when multiple large coding jobs are being run on the same day.
ChatGPT Work Also Faces The Restriction
The five-hour limit also applies to ChatGPT Work for Plus subscribers.
ChatGPT Work is intended for longer research and deliverable-oriented tasks, making it another type of workload where usage can accumulate quickly. OpenAI’s current documentation describes ChatGPT Work as a separate experience for longer research and deliverables, while Codex is focused on coding-agent tasks.
The combined return of the restriction reflects OpenAI’s attempt to manage a broader category of agentic workloads rather than ordinary text-chat volume alone.
What Happens When A User Hits A Limit?
Once a user reaches an applicable five-hour or weekly usage limit, available options can include waiting for the relevant allowance to reset or purchasing additional credits where that option is available.
| Situation | Likely User Action |
|---|---|
| Five-hour limit reached | Wait for the usage window to reset |
| Weekly allowance reached | Wait for the weekly reset |
| Additional credits available | Purchase credits for continued usage |
| Heavy recurring workload | Consider a higher-tier plan |
| Occasional usage | Continue using Plus within available limits |
OpenAI also periodically provides usage resets or promotions, although these should not be treated as a permanent part of the Plus allowance.
Pro Users Are Not Affected For Now
One of the most notable aspects of the policy change is the difference between Plus and Pro.
OpenAI said the five-hour restriction will remain disabled for its $100 and $200 Pro subscriptions for the upcoming months. That creates a clearer distinction between Plus, which is aimed at a broader group of users, and Pro, which is designed for people with heavier AI workloads.
The current structure can therefore be summarized as:
CHATGPT USAGE ACCESS
Plus
├── ChatGPT Work → 5-hour limit restored
├── Codex → 5-hour limit restored
└── Weekly allowance → Still applies
Pro $100 / $200
├── 5-hour limit → Disabled for now
└── Weekly allowance → Still applies
Enterprise / Edu
└── Separate credit-based system
Sottiaux indicated that the exemption for Pro is planned to continue for the coming months, although future changes remain possible.
The Policy Comes During A Period Of Rapid ChatGPT Expansion
The reinstatement comes amid rapid changes to OpenAI’s product and usage structure.
Earlier in August, OpenAI announced broader access to GPT-5.6 models and expanded unlimited text-chat access for Free users, while maintaining separate limits for features such as files, images, voice and image generation. Plus and Pro subscribers also received an upgraded GPT-5.6 Sol experience.
That expansion increases the importance of capacity management. As more users access more capable models and agentic features, OpenAI must balance availability, infrastructure costs and the predictability of paid-plan usage.
The five-hour cap is therefore part of a larger trend in AI services toward differentiated access based on workload intensity rather than simply the number of messages a user sends.
What The Change Means For OpenAI’s Business Model
The policy also highlights the economics of generative AI.
Traditional software subscriptions can often support additional users at relatively low marginal cost. AI workloads are different because every model interaction requires computing resources, and agentic tasks can generate considerably more inference and tool-use activity.
A five-hour limit gives OpenAI a mechanism to control periods of unusually high demand without necessarily reducing the total weekly allowance. At the same time, higher-tier plans provide a pathway for users who need greater access.
This creates a tiered structure:
Casual users → Plus → Heavy users → Pro → Enterprise
The strategy allows OpenAI to offer broad access while charging more for customers whose workloads require greater capacity.
The Bigger Picture
The return of the five-hour limit shows that AI subscription plans are increasingly being designed around compute consumption and workload intensity, not simply traditional message counts. Codex and ChatGPT Work can involve sustained, resource-intensive operations, making short-term capacity controls useful for OpenAI as demand grows.
For Plus users, the immediate effect is a reduction in flexibility for intensive Work and Codex sessions. However, OpenAI is retaining the broader weekly allowance and has left the restriction disabled for Pro users for now. The change therefore creates a sharper distinction between mainstream paid access and premium access aimed at heavy AI users.
Looking Ahead
The key issue for Plus subscribers will be whether the restored five-hour window provides enough flexibility for real-world coding and research workflows. Users who routinely conduct long sessions may need to spread tasks across the day or week, while those who repeatedly encounter the cap may find the higher-priced Pro tiers more attractive. OpenAI’s decision to leave the five-hour restriction disabled for Pro also gives the company a way to observe demand before deciding whether the same control should eventually apply to heavier plans.
For OpenAI, the challenge is balancing customer expectations with the rapidly increasing cost of advanced AI workloads. If model capabilities and agentic features continue expanding, usage limits are likely to become an increasingly important part of subscription design. The reinstated Plus limit is therefore more than a temporary policy reversal: it offers a glimpse into how AI companies are managing the economics of increasingly powerful and compute-intensive products.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
