OpenAI will combine ChatGPT and Codex subscription usage allowances

OpenAI says that starting in December, ChatGPT and Codex on Plus, Pro, and Business plans will draw from one shared allowance. The new meter is tokens rather than two separate product limits. OpenAI says Instant chat will not consume the shared allowance.

This is a subscription and capacity change across products, not a feature update to either one. A complex Codex coding task or a non-Instant ChatGPT conversation can reduce the capacity available in the other product on the same plan. Consumption also varies with the model and task, so a prior intuition based on requests or separate limits may not predict the new remaining balance.

OpenAI says Settings → Usage will show the combined remaining allowance and reset times. Once a subscriber reaches the limit, OpenAI says they can wait for a reset, upgrade, or buy credits for eligible features. The help article expects most users not to wait for resets more often than today, but it does not state fixed token limits for each plan or a uniform token cost per task.

Teams that use both products should record high-consumption coding-agent work, long-context analysis, and ordinary Instant chat separately before the transition, then compare actual token use and reset behavior in December. Do not infer future capacity from today's separate meters; use the account's live usage view and OpenAI's current documentation.

Related directory entries: ChatGPT · Codex CLI