OpenAI to Share ChatGPT and Codex Usage Quotas Starting December 2026

OpenAI announced on its official help page that starting in December 2026, the usage allowances for ChatGPT Plus, Pro, and Business will be shared between ChatGPT and the coding agent Codex.

Understanding ChatGPT and Codex usage limits - OpenAI Help Center

ChatGPT offers two services that users can use differently: Chat, which provides conversational assistance, and Work, which handles tasks requiring multiple steps such as research and document creation. OpenAI has indicated a policy of integrating Work's agent functionality into regular ChatGPT, and this sharing of usage allowances accompanies that integration. Note that usage of ChatGPT Instant mode, used for everyday conversation, is not counted toward consumption of the shared allowance.

Usage is measured in tokens, the unit of information that the model processes and generates. How much of the shared allowance is consumed varies depending on the model selected and the task requested, with a tendency for advanced models and complex tasks to consume more. The remaining credit balance and reset time will be viewable under "Settings" → "Usage" in the menu.

Expanding Work and Codex usage allowances for Pro and Business premium users

Ahead of the December sharing, OpenAI announced on October 29 an expansion of the Work and Codex usage allowances included in Pro and Business premium seats.

The changes to Pro usage allowances are as follows.

Pro 100 / Pro 500: Work and Codex usage allowances doubled from before. Pro 200: The reduction announced in the September 29 email is withdrawn; contracts that already had the larger allowance applied will keep that allowance, contracts that had the smaller allowance applied will have their allowance doubled, and all Pro 200 contracts are aligned to the same cap.

Limits on the number of contracts per device or payment card

Separately from the sharing of usage allowances, OpenAI also provided guidance on limits to the number of contracts tied to the same payment card or device.

Understanding card and device limits in ChatGPT - OpenAI Help Center

If too many contracts are tied to the same card or device, new purchases may become impossible, and an upper limit is also set on the number of paid accounts that can be used on a single device. When the usage limit is triggered, "Device limit reached" is displayed, and the target paid account can no longer be used on that device. Even if the account was usable immediately after purchase, a subsequent check may sign the user out. The specific upper limit numbers and the start date for applying this restriction are not disclosed. Additionally, for contracts purchased before the restriction was applied, there are cases where they can continue as existing contracts. However, it is not guaranteed that a new contract can be made after cancellation.

The device restriction applies to ChatGPT Go, Plus, Pro 100, Pro 200, and Pro 500. Free accounts and the Business, Enterprise, and Edu workspace plans are not included in this restriction.

Note that while it is possible to hold more than two accounts, the account switching feature can keep at most two accounts signed in at the same time. Even when using multiple accounts, contracts and usage allowances are separate for each account. Also, because the number of accounts that can be signed in simultaneously and the number of paid accounts usable on a single device are subject to separate limits, signing out once does not necessarily mean another paid account can be used on that device.

For this reason, even if family members etc. each use their own account, they may reach the limit if they share a card or device. This restriction is not lifted by purchasing credits or resetting usage allowances. The company advises users who believe the restriction was applied in error to check with support.

If a target paid account becomes unusable due to the device restriction, cancellation of the contract and a refund can be requested. The procedure differs depending on where the purchase was made. For specific procedures and refund details, refer to OpenAI's help.