OpenAI will merge the usage allowances on its paid ChatGPT plans into a single pool from December, so chatting in ChatGPT and coding in Codex will draw from the same balance. The change covers Plus, Pro and Business subscribers, and usage will be measured in tokens rather than separate limits. Instant chat, the fast default mode, will stay outside the meter. OpenAI says most users should not need to wait for a reset more often than they do now, but it has not published an exact start date or the size of any plan’s allowance.
What changes in December
Today, ChatGPT and Codex carry separate allowances. From December, a Plus, Pro or Business plan will have one shared allowance across both products, OpenAI says in its help article on the change, updated on October 31. The company presents the move as part of bringing ChatGPT Work’s agentic capabilities into the main ChatGPT experience.
Measurement changes with the merger. Usage will be counted in tokens, the units of information a model processes and generates, so how much of the allowance a request consumes will depend on the model chosen and the task. More advanced models and more complex tasks will generally use more of the allowance, OpenAI says.
One mode escapes the meter entirely. Instant chat, the fast default model for quick questions, will not draw from the shared allowance at all. Subscribers will see combined remaining usage and reset times in Settings, under Usage. When the allowance runs out, they can wait for a reset, upgrade, or buy credits.
The change also clarifies what already shares a pool. On Plus and Pro, Codex, ChatGPT Work, Excel and PowerPoint already share one agentic allowance when those features are available on the plan. December mainly pulls everyday chat with advanced models into that same pool, leaving Instant chat routed around it.
Why the timing matters
The help article did not arrive out of nowhere. Two days earlier, OpenAI doubled Work and Codex usage for Pro subscribers and cancelled a planned reduction for one Pro tier. The sequence suggests OpenAI knew the merger would land hardest on people who use both products heavily, and moved first to give them headroom.
Sharp-eyed users had spotted the change coming. On October 29, a checkout-page line warned that plan usage would share a single ceiling across ChatGPT and Codex from December, before the help article confirmed it. OpenAI has still not published the allowance size for any plan, so nobody outside the company can yet say whether a given subscriber gains or loses.
What OpenAI has not said matters as much as what it has. There is no exact December date, no per-plan allowance size, and no conversion table showing how chat tokens and Codex tokens count against the same balance. Until those numbers appear, subscribers cannot plan precisely, and every estimate of the change’s cost is guesswork dressed as arithmetic.
Our read: one meter, sharper trade-offs
The honest way to describe this change is that it prices ChatGPT use by what each request costs OpenAI, not by which product window it happens in. That is fairer in one sense: a subscriber who chats lightly and codes heavily was previously sheltered by two separate buckets, and now pays for total consumption instead.
But the merger also removes a useful safety valve. A long coding afternoon can now eat into the balance a subscriber also needs for everyday chat, and advanced models drain the pool faster. Developers who live in Codex all day will feel that first, while casual subscribers who mostly use Instant chat may notice nothing at all.
The contrast with the funding climate is hard to ignore. OpenAI has just trimmed its latest fundraising target, and a metering change that ties revenue more closely to compute cost looks like part of the same discipline. Token metering is also where the industry is heading, and OpenAI is not moving alone.
Why It Matters
For subscribers, the practical step is simple and immediate: start watching Settings, under Usage, now, before December arrives, so the first shared-allowance month does not come as a surprise. Heavy users should measure their own burn across chat and Codex separately while they still can, because once the pools merge, that history disappears into one number.
The move confirms flat AI allowances are giving way to metered consumption, with each lab choosing its own unit. Microsoft is moving Copilot Business the same way, with credits rather than tokens, and Anthropic has been testing larger and smaller budgets on its Max plans. Headline price alone will increasingly miss the real question: how far a plan goes.
Background
Codex began as OpenAI’s coding agent and has grown into a part of its paid plans, alongside ChatGPT Work, a set of agent features for spreadsheets, documents and presentations. Work and Codex already shared one agentic allowance on Plus and Pro, so December’s change completes a consolidation OpenAI has been building toward all year, pulling chat into the same accounting.
Source: OpenAI’s help article
0 Comments