Skip to content
Hire an AI employee
AI EmployeeSuper-AgentsAgent-to-AgentBenchmarksPricingBlog
Guides 08 Choosing a platform

ChatGPT and Codex Will Share One Usage Limit in December

Nitish Garg · Founder & CEO, CellCog · 01 November 2026 · 6 min read · 1,216 words
At a glanceQuick answers
What changes?
From December, Plus, Pro and Business get one allowance for ChatGPT and Codex together, measured in tokens.
Is anything exempt?
Yes. Instant chat will not draw from the shared allowance.
What happens at the limit?
You wait for the reset, upgrade your plan, or buy credits, as OpenAI’s help page lists.
Editorial data illustration on a near-white ground: ChatGPT and Codex gauges pipe into one navy and teal tank labelled One allowance, from December, beside a large amber 1 shared limit and two token streams, a long one from an advanced model and a short one from a smaller model
Fig 0Two meters, one allowance. Made by CellCog's image agent, running GPT Image 2.5.

OpenAI will pool the usage on ChatGPT’s paid plans. Starting in December, Plus, Pro and Business subscribers get one allowance that covers both ChatGPT and Codex, measured in tokens instead of separate limits. The change sits in a help-center article, Understanding ChatGPT and Codex usage limits, updated on October 31, 2026. We found no OpenAI blog post or X post announcing it.

The article opens with the scope: “Starting in December, Plus, Pro, and Business plans’ usage allowance will be shared across ChatGPT and Codex as we bring the agentic capabilities of Work into the main ChatGPT experience.” Users spotted the page on X the same afternoon, and one post summing it up passed 89,000 views within six hours.

On this page · 8 sectionsOpen
  1. What changes in December
  2. What already shares, and what joins it
  3. How much a token costs depends on the model
  4. What OpenAI has not said
  5. How heavy users took it
  6. Our read
  7. What we are watching
  8. Update log
Key points6 · 6 min full read
  1. A chat bubble and a code bracket feeding one gauge.
    From December, Plus, Pro and Business plans will share one usage allowance across ChatGPT and Codex, per OpenAI’s help center.
  2. A row of tokens flowing into a meter.
    Usage will be measured in tokens, so more advanced models and harder tasks use more of the allowance.
  3. A lightning bolt in a chat bubble with an infinity sign.
    Instant chat will not draw from the allowance.
  4. Three bars of rising height, the tallest amber.
    On Plus and Pro, Work, Codex, Excel and PowerPoint already share one agentic allowance; December adds chat with the advanced models.
  5. A settings gear beside a progress bar.
    Combined remaining usage and reset times will show in Settings, and at the cap you wait, upgrade or buy credits.
  6. A calendar page with a question mark.
    OpenAI gives no exact date and no allowance sizes, and we found no blog or X post announcing the change.

§ 01What changes in December

Animated diagram: separate Chat and Codex and Work meters pipe into one tank labelled One allowance, in tokens; token dots drain in a thick stream for an advanced model and a thin one for a smaller model, while an amber Instant chat bubble's dashed arc goes around the tank, labelled Does not count
Fig 1Two meters become one allowance in tokens; Instant chat goes around it
Question Answer from OpenAI’s help page
Which plans Plus, Pro and Business
What is shared One usage allowance across ChatGPT and Codex
How it is measured Tokens, the units a model reads and writes
What is exempt Instant chat
Where to check Settings, Usage: combined remaining usage and reset times
At the limit Wait for the reset, upgrade the plan, or buy credits
When December, no exact date
Table 1What OpenAI’s help page says changes (read November 1, 2026)

“Your usage for ChatGPT and Codex will be measured in tokens, the units of information a model processes and generates.” That makes the cost of a request depend on what it asks for: “More advanced models and complex tasks generally use more of your allowance.” The one exception is the fast default model: “Instant chat will not draw from your usage.”

OpenAI also tries to settle nerves: “We expect most users won’t need to wait for their usage to reset more often than they do now.” When a user does run out, “you can wait for it to reset, upgrade your plan, or buy credits to continue using eligible features.”

Three plan badges, Plus, Pro and Business, feed one teal bar labelled ChatGPT plus Codex measured in tokens, with Instant chat routed around it

01One shared allowance for Plus, Pro and Business; Instant chat does not count

Two navy boxes for chat with advanced models and for Codex, Work, Word, Excel and PowerPoint merge into one teal box labelled One shared allowance

02Today two meters, from December one allowance for chat and agent work

Bar chart of output credits per million tokens: GPT-5.6 Luna 30, Terra 300, Sol 500 and an amber GPT-6 Astra bar at 1,250, with a large 50x

03On the Office rate card, GPT-6 Astra output costs 50 times GPT-5.6 Luna

1 / 3
Fig 2What changes in December, in three slides

§ 02What already shares, and what joins it

On personal plans, the pooling started before this. OpenAI’s credits page already says: “For Plus and Pro, Codex, ChatGPT Work, Excel, and PowerPoint can share the same agentic usage allowance when those features are available on your plan.” December folds chat with the advanced models into that pool, and leaves Instant out of it. Business seats on credit-based pricing already pay token rates for Work and Codex under OpenAI’s rate card.

Use Today From December
Codex Shared agentic allowance One shared allowance
ChatGPT Work, Word, Excel, PowerPoint Shared agentic allowance One shared allowance
Chat with advanced models Separate chat limits One shared allowance
Instant chat Separate Does not count
Table 2Before and after on Plus and Pro, per OpenAI’s help pages

§ 03How much a token costs depends on the model

OpenAI has not said how the shared allowance converts tokens. The closest published guide is the rate card for its Office features, which already bill by the token on personal plans. There, output from GPT-6 Astra costs 1,250 credits per million tokens, 50 times GPT-5.6 Luna’s 30.

Model Input Cached input Output
GPT-5.6 Luna 5 0.5 30
GPT-5.6 Terra 50 5 300
GPT-5.6 Sol 100 10 500
GPT-6 Astra 250 25 1,250
Table 3Credits per 1 million tokens, ChatGPT for Word, Excel and PowerPoint (OpenAI’s credits page, read November 1, 2026)
Output credits per 1M tokens: GPT-6 Astra costs 50 times GPT-5.6 LunaBar chart of output credits per million tokens in OpenAI's Office rate card: GPT-5.6 Luna 30, GPT-5.6 Terra 300, GPT-5.6 Sol 500, GPT-6 Astra 1,250 highlightedGPT-5.6 Luna30GPT-5.6 Terra300GPT-5.6 Sol500GPT-6 Astra1250Output credits per 1M tokens: GPT-6 Astra costs 50 times GPT-5.6 LunaBar chart of output credits per million tokens in OpenAI's Office rate card: GPT-5.6 Luna 30, GPT-5.6 Terra 300, GPT-5.6 Sol 500, GPT-6 Astra 1,250 highlightedGPT-5.6 Luna30GPT-5.6 Terra300GPT-5.6 Sol500GPT-6 Astra1250
Fig 3Output credits per 1M tokens: GPT-6 Astra costs 50 times GPT-5.6 Luna

Read it as the shape of the spread, not the December math: these rates cover the Office features, and OpenAI has not published the size of the shared allowance on any plan.

§ 04What OpenAI has not said

  • A date. The page says December and nothing more precise.
  • Allowance sizes. No plan gets a number, and the page does not say whether the size changes.
  • The conversion. How chat tokens and Codex tokens count against the same allowance is not on the page.
  • Business details. Business is named, with no word on how seats on credit-based pricing change.

§ 05How heavy users took it

The first reactions came from people who use Codex all day. VraserX argued for keeping the limits apart: “Heavy Codex usage shouldn’t eat into your ChatGPT allowance.” LuminaBench went further: “if they make this feel terrible Anthropic can happily take my money”. The worry in both is the same: a long coding run now draws down the same balance as everyday chat.

§ 06Our read

Two days earlier OpenAI doubled Work and Codex usage on Pro; now it moves every paid plan to one pool, priced by what each request uses. Anthropic has been testing larger and smaller budgets on Claude Max at the same time. Token metering ties a plan closer to what each request costs the lab, and it means the model you pick now decides how far your plan goes.

CellCog’s AI employees run on Claude Opus 5.5. On CellCog you pay for the work, not the hire, and the cost depends purely on how much work you assign.

§ 07What we are watching

  • A date and allowance sizes for each plan.
  • A token rate card for chat models on personal plans.
  • Whether the October 29 Pro doubling carries into the shared allowance.

§ 08Update log

As of November 1, 2026, 01:15 UTC: page opened from OpenAI’s help article (updated October 31), its credits and rate card pages, and reactions on X.

Frequently asked5 questions

Q1When does the shared limit start?

OpenAI’s help page says December 2026 and gives no exact date.

Q2Which plans are affected?

Plus, Pro and Business. The help page does not mention Free, Go or Enterprise.

Q3Does every chat count against it?

No. Instant chat will not draw from the allowance. Chats with more advanced models will, and they use more of it.

Q4Will I hit the limit more often?

OpenAI says it expects most users won’t need to wait for a reset more often than they do now. It has not published allowance sizes.

Q5Where can I see my usage?

OpenAI says combined remaining usage and reset times will show in Settings, under Usage.

Published 01 November 2026 All Choosing a platform →