OpenAI is changing how subscribers use ChatGPT and Codex by introducing one shared, token-based usage allowance starting in December 2026. The change applies to Plus, Pro and Business plans, meaning activity in either product will draw from the same allowance rather than having separate usage limits.
The company says consumption will depend on the model selected and the complexity of the task. More advanced models and demanding tasks generally use more of the allowance, while Instant chat will not count towards the shared usage pool. Users will be able to monitor their combined remaining allowance and reset times under Settings → Usage.
Key Takeaways
- OpenAI will introduce shared ChatGPT and Codex usage limits starting in December 2026.
- The change covers Plus, Pro and Business plans.
- Usage will be measured in tokens rather than a fixed number of messages.
- Both ChatGPT and Codex activity will consume the same allowance.
- Instant chat will remain outside the shared usage limit.
- More advanced models and complicated tasks generally consume more allowance.
- Users can check remaining usage and reset times in Settings → Usage.
- OpenAI says most users should not need to wait for their allowance to reset more frequently than they do now.
- Subscribers who reach their limits may wait for a reset, upgrade their plan or purchase credits for eligible features.
What Is Changing in ChatGPT and Codex?
Until now, ChatGPT and Codex have had separate usage arrangements.
ChatGPT is primarily used for conversations, research, writing, analysis and other AI-assisted activities. Codex is OpenAI’s coding agent, designed to help developers understand codebases, write software, fix bugs and complete development tasks.
OpenAI is moving towards a shared usage system in which both products draw from one allowance.
This means a subscriber who uses a significant amount of Codex for software development will also have less of the shared allowance available for eligible ChatGPT and Work activities during the same usage period.
The reverse is also true. Heavy use of eligible ChatGPT models can reduce the allowance available for Codex.
The change is intended to bring the products closer together as OpenAI expands its agentic capabilities and integrates Work more closely into the main ChatGPT experience.
How Will the Shared Token Allowance Work?
The new system will measure usage in tokens.
Tokens are the units of information that AI models process and generate. They can represent parts of words, complete words, punctuation and other information.
Under OpenAI’s approach, the amount of allowance consumed will depend on several factors.
| Factor | Effect on usage |
|---|---|
| Model selection | More advanced models generally consume more allowance |
| Task complexity | Complicated tasks generally require more usage |
| Input size | Larger amounts of information can increase consumption |
| Output requirements | Longer or more demanding responses can use more resources |
| Agent activity | Multi-step tasks can require more model processing |
This means the number of messages a subscriber can send will not be fixed.
A short question using a lightweight model may consume relatively little allowance, while a complicated coding task involving a large project and several steps may use substantially more.
OpenAI has not published a universal token allowance for each plan in the announcement. Subscribers will need to check their account’s usage dashboard for their available allowance and reset information.
Which ChatGPT Plans Are Affected?
The shared usage system is scheduled to begin in December 2026 for three subscription categories.
| ChatGPT plan | Shared ChatGPT and Codex allowance |
|---|---|
| Plus | Yes |
| Pro | Yes |
| Business | Yes |
| Free and Go | Not included in this announced change |
The announcement specifically identifies Plus, Pro and Business plans.
It does not provide a detailed breakdown of the exact shared allowance for every subscription tier. Users should therefore avoid assuming that all three plans will receive the same amount of usage.
The allowance available to an individual subscriber will depend on their plan and the applicable usage rules.
Instant Chat Will Not Consume the Shared Allowance
One important exception is Instant chat.
OpenAI says Instant chat will not draw from the shared usage allowance.
This means users can continue using the Instant experience without consuming the shared ChatGPT and Codex pool described in the announcement.
The distinction matters because different models have different resource requirements.
More advanced models are designed for demanding reasoning, coding and research tasks, while faster models can handle many everyday questions.
Keeping Instant chat outside the shared allowance could help subscribers preserve their more limited usage capacity for tasks that require advanced models.
What Happens When Users Reach Their Limit?
OpenAI says users will be able to see their combined remaining usage and reset times through:
Settings → Usage
If a subscriber reaches the shared limit, the company identifies three possible options:
- Wait for a reset: Users can continue once their applicable usage allowance refreshes.
- Upgrade the plan: Subscribers may move to a plan with a higher allowance, subject to availability.
- Purchase credits: Eligible features can be accessed using additional credits where supported.
OpenAI also says it expects most users will not need to wait for their usage to reset more frequently than they do currently.
However, this does not mean that every subscriber will have exactly the same experience. People who regularly use advanced models or run lengthy coding agents may consume their allowance faster than casual users.
Why Is OpenAI Combining ChatGPT and Codex?
The change is connected to OpenAI’s broader effort to make ChatGPT a platform for completing tasks rather than simply answering questions.
The company is bringing its agentic Work capabilities closer to the main ChatGPT experience.
Work is designed for more complicated activities, including research, analysing files, navigating supported websites and producing finished documents.
Codex focuses on software development, allowing users to work with codebases and delegate programming tasks.
Combining the products under a shared usage system reflects their growing similarities: both can involve extended interactions, multiple steps and substantial model processing.
A unified allowance could also make it easier for OpenAI to manage how computing resources are allocated across its products.
However, it reduces the independence users previously had when managing their ChatGPT and Codex usage.
What Does This Mean for Heavy Users?
The impact will depend largely on how subscribers divide their time between ChatGPT and Codex.
For someone who mainly uses ChatGPT for everyday questions, the change may have limited practical impact.
For a developer who regularly runs Codex agents while also using advanced ChatGPT models for research and planning, the shared allowance could become more important.
Consider a developer working on a software project. They might use ChatGPT to plan the architecture, analyse technical documentation and investigate errors before switching to Codex to implement changes and run tests.
Under the new arrangement, eligible usage across both products will count against the same allowance.
Users may therefore need to prioritise advanced models for difficult tasks and use more economical models for routine work.
The Bigger Picture
OpenAI’s decision reflects a broader change in the AI industry. Companies are increasingly moving from standalone chatbots towards AI platforms that combine research, coding, automation and multi-step task execution.
These capabilities require different amounts of computing resources. A simple conversation and an autonomous coding session can have very different processing requirements.
A token-based shared allowance gives OpenAI a way to measure consumption across multiple products rather than maintaining completely separate usage systems.
For subscribers, however, the key question will be whether the combined allowance provides enough capacity for both everyday conversations and demanding agentic workflows.
Looking Ahead
The shared allowance is scheduled to begin in December 2026. OpenAI will need to communicate the plan-specific limits clearly so subscribers can understand how much usage they receive and how different models and tasks affect their remaining capacity. The usage dashboard will become particularly important for people who regularly switch between ChatGPT and Codex.
The longer-term impact will depend on how OpenAI balances its expanding agent capabilities with subscription limits. A unified system could make its products easier to manage, but heavy users will want transparency about consumption, reset periods and the amount of work included in each paid plan.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
