OpenAI puts ChatGPT and Codex on one shared allowance from December

OpenAI puts ChatGPT and Codex on one shared allowance from December

31 October 2026: OpenAI will make ChatGPT and Codex draw on one usage allowance for Plus, Pro and Business plans starting in December, measured in tokens, as it folds the agentic features it calls Work into the main ChatGPT experience. Until now chat and Codex usage were metered separately, so a team that spent its coding quota could still use chat, and the other way round.

Not yet independently verified. This rests on one short OpenAI help article, updated on 31 October, which gives "December" but no exact date and no allowance sizes per plan. We have not yet read the customer email OpenAI sent about the change. We will update this when it can be confirmed, and remove this note.

Tokens instead of message counts

The help article says usage for both products "will be measured in tokens, the units of information a model processes and generates." How much of the allowance a request uses will depend on the model chosen and the task: more advanced models and complex tasks use more. One part of the product stays outside the budget, since "Instant chat will not draw from your usage."

That makes the cost of a request depend on its size rather than on the number of prompts. A long reasoning session with attached files, a deep research run or an agentic coding task in Codex will all come out of the same pool, and the pool resets on a schedule the user will see in Settings, under Usage, with the combined remaining usage and the reset times.

What happens at the limit

OpenAI lists three options for a user who reaches the shared limit: wait for the reset, upgrade the plan, or buy credits to keep using eligible features. It says it expects most users "won't need to wait for their usage to reset more often than they do now", but it does not publish the size of the allowance for any plan, so that expectation cannot yet be checked against current limits.

For Business workspaces the practical question is who runs out first. Developers who spend most of the day in Codex will now use up the same budget they rely on for ordinary chat, and colleagues who mostly chat will carry unused capacity they cannot hand over. Whether credits are bought per seat or for the workspace is not stated in the article.

Planning for the change

Teams that bought Pro seats mainly for chat with the strongest models, and used Codex on the side, are the ones most likely to notice. The reaction on developer forums after the article was updated came largely from people who had chosen ChatGPT because heavy chat use did not eat into coding time.

Before December, an administrator can note each heavy user's current pattern, so that the first month under the shared budget can be compared with something. Our AI coding assistant category lists six products for teams weighing an alternative for the coding half of the work.

Related on TrustList:

• Copilot Studio app leaves Teams, editing ends in January 2027

Sources

• OpenAI Help Center: Understanding ChatGPT and Codex usage limits (updated 31 October, read 1 November), 31 October 2026