ChatGPT and Codex will share token-based limits starting in December
ChatGPT Plus, Pro and Business users will share a token-based allowance across ChatGPT and Codex starting in December, according to OpenAI’s usage-limit guidance. Instant chat will be exempt. This is a planned transition, not a change readers should assume has already reached their accounts.
What the December change covers
OpenAI connects the change to bringing Work’s agentic capabilities into the main ChatGPT experience. Consumption will vary with the model and task, rather than treating every request equally. The company expects most users will not face more frequent waits for resets, but that is a forecast, not a guaranteed allowance.
The guidance does not specify token totals, an exact December start date or a plan-by-plan conversion. It does not establish December terms for plans outside the three named tiers.
Shared usage is not entirely new
OpenAI’s current Codex plan documentation already describes a shared allowance and credit pool for Codex, ChatGPT Work, ChatGPT for Excel and Workspace Agents on applicable plans. It also says task complexity, context, reasoning, speed and tools affect consumption. A long-running task can therefore use more of that existing allowance than a short request.
That distinction matters when comparing the announcement with today’s account dashboard. Existing shared agentic usage and the planned broader ChatGPT transition are not interchangeable descriptions of the same rollout.
The current Work and Codex pricing guide also separates API token prices from subscription usage. It explicitly warns against using API rates to estimate included tasks. Token-based accounting does not mean an ordinary subscriber is automatically billed at the developer API rate.
This planned transition is separate from the October Work and Codex allowance changes.
What to check in your account
For current Codex limits, the existing documentation directs users to Settings or the usage dashboard to identify the exhausted allowance, credit balance and reset time. In an active Codex CLI session, /status provides usage information. Follow the options shown for the account rather than assuming every plan offers the same remedy.
The announcement does not yet support calculating how a particular mixed chat-and-coding workload’s capacity will change.
Editorial illustration by ByteForward, generated with AI. OpenAI wordmark: OpenAI. Not a product screenshot or quota diagram.
