- From December, OpenAI says Plus, Pro and Business plans will share one token allowance across ChatGPT and Codex. Instant chat does not draw on it.
- The flat-rate promise is narrowed, not removed: Instant chat stays outside the meter, while model choice and task complexity decide how fast the rest is spent.
- Leaders should decide who picks models, who watches usage, what happens at the limit, and ask OpenAI how Business allowances are assigned.
A flat fee with a meter attached
A support page OpenAI updated describes a December change for three paid plans: Plus, Pro and Business. Today ChatGPT and Codex are separate products. After the change, one pool of usage will cover both. Instant chat sits outside that pool. The plan therefore has two parts: a tier that stays outside the meter, and a tier where every choice of model spends something.
OpenAI gives a reason: it is folding the agentic features of Work into the main ChatGPT experience. An agent, in plain terms, is software that works through a task for you instead of answering once. The Decoder points out that the Chat and Work merger had been announced before this page appeared.
What is metered and what is not
OpenAI says it will count usage in tokens, which it describes as "the units of information a model processes and generates." That wording covers both directions. What goes in and what comes out both spend allowance, so long prompts and long outputs both count.
OpenAI names two drivers of spend: which model you pick and what task you give it. Stronger models and harder tasks generally use more of the allowance.
The Decoder puts the same point differently. In its words, complex models and tasks "eat through the budget faster." Chats with Instant models do not count toward the budget.
The page does not list which conversations count as Instant chat. It also gives no token totals per plan and no conversion between chats and agent tasks. WebPulse cannot say how far any allowance will stretch.
Why the split matters
This section is interpretation, not OpenAI's claim. The lesson here is that the flat-rate promise has been narrowed, not removed. Some use stays free of the meter. Everything else becomes a series of small spending decisions.
The Decoder says ChatGPT stays a flat-rate service "on paper" because Instant chats are exempt. It also predicts that cheaper plans will deliver weaker output under heavy use. That is the outlet's forecast. OpenAI does not state it, and WebPulse cannot confirm it.
What the source does support is simpler. From December, picking a stronger model will cost allowance, and so will handing work to an agent. For one person that is a habit. For a company it is a policy, whether or not anyone writes it down.
What users will see, and what is not said
OpenAI says Settings → Usage is where each person can check what remains and when it resets. The page does not say whether Business administrators can see individual use. It also does not say how Business allowances are divided across people.
At the limit, OpenAI lists three options: wait for the reset, upgrade the plan, or buy credits to keep using eligible features.
OpenAI says it is preparing carefully to avoid disrupting workflows. It expects most users will not wait for resets more often than they do now. That is an expectation, not a published limit.
What The Decoder adds
The Decoder reports that OpenAI recently doubled usage limits on higher-priced plans. It suggests this was preparation for the merger. OpenAI's page does not mention the increase.
Today's ChatGPT offers two operating modes and 11 model settings, according to The Decoder. It reads the merger as a likely answer to criticism of that growing complexity. It also notes that Anthropic has already combined Chat and Cowork in Claude.
Questions to put to your team
First, which of these plans do we hold? OpenAI says the change applies to Plus, Pro and Business.
Second, who decides which model is used for everyday work? If nobody does, individual habits will set how fast allowances run down.
Third, who checks Settings → Usage, and how often? Reset times only help if someone looks before a deadline.
Fourth, ask OpenAI how Business allowances are assigned and whether administrators can see usage. The support page leaves both open.
Fifth, what is our rule at the limit? Waiting, upgrading and buying credits carry different costs. Set the policy before the day it is needed.
Before December, list the tasks your staff run most. Once the change takes effect, measure how much allowance they use.
A flat fee asks nothing of its owner. A measured allowance asks someone to keep watching the gauge.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: OpenAI.
