ChatGPT and Codex will share one usage limit from December. What changes

What OpenAI confirmed on October 31

On October 31, 2026, OpenAI published a help center article, Understanding ChatGPT and Codex usage limits. It opens: “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.”

The rest of the article makes four points:

• Usage is counted in tokens. “How much of your allowance you use will depend on the model you choose and the task you ask it to do. More advanced models and complex tasks generally use more of your allowance.”
• Instant is left out. “Instant chat will not draw from your usage.”
• One meter. “You will be able to see your combined remaining usage and reset times in Settings → Usage.”
• The same ways out. “If you reach your limit, you can wait for it to reset, upgrade your plan, or buy credits to continue using eligible features.”

On how often people will run out, OpenAI says: “We expect most users won’t need to wait for their usage to reset more often than they do now.”

Subscribers got an email as well. One who posted on Hacker News said it announced that “Chat and Work are coming together in one experience,” and that halfway through came the real headline: “A shared allowance for ChatGPT and Codex.” We haven’t seen the email ourselves. OpenAI’s release notes hadn’t listed the change by the morning of November 1; their latest entry was from October 30.

What counts against which limit

Today, Codex shares one allowance with OpenAI’s agent features: “Codex, ChatGPT Work, ChatGPT for Excel, and Workspace Agents use a shared allowance and credit pool when those features are available on your plan,” says its Codex help article. Chat keeps its own. “Usage and credit rules in Work and Codex are separate from Chat,” says the models article. December changes that on three plans:

What you use	Until December	From December, on Plus, Pro and Business
Codex, in the app, CLI, IDE extension or cloud	The Work and Codex allowance	The shared ChatGPT and Codex allowance
ChatGPT Work	The Work and Codex allowance	The shared ChatGPT and Codex allowance
Instant chats	Unlimited, “subject to abuse guardrails”	Not counted
Thinking in Chat, Medium to Extra High	Chat’s own Thinking limits	Not said
GPT-6 Pro in Chat	Its own weekly message limit	Not said
Voice, images and file uploads in Chat	Their own limits	Not said
Your dot	Its own	Not said

The “Until December” column comes from OpenAI’s pricing page, which lists everyday text chats as “Unlimited” on Free, Go, Plus and Pro; the models article, which says “Thinking limits depend on your plan”; the GPT-6 Astra article, which says GPT-6 Pro’s “message limits are separate from the shared Work and Codex allowance”; the Codex help article, which says “ChatGPT file uploads, image generation, and voice have separate usage limits, reset periods, and banners”; and OpenAI’s October 29 statement, as BleepingComputer quotes it, that “usage of your dot will still continue to be separate from your default usage.”

OpenAI exempts only Instant by name, and says what you use depends on “the model you choose,” which reads as though every other chat draws on the shared allowance. It hasn’t said so row by row, so the table says “Not said.”

Codex already counts this way, and its pricing page shows how far apart the models are. In credits per million tokens, GPT-6 Astra, which powers GPT-6 Pro, costs 250 for input and 1,250 for output, five times GPT-6 Sol’s 50 and 250. Sol powers Chat on paid plans, from Instant to Extra High. OpenAI hasn’t said chats will be counted at these rates, and notes that “credit prices alone don’t determine included subscription usage.”

Who it affects

• Plus. Codex and Work already share an allowance with a five-hour window: Codex’s pricing gives “local messages per five-hour period for Plus and Standard Business,” and “weekly limits may also apply.” From December, ChatGPT shares it too, Instant chats aside, so a long afternoon of chats above Instant can leave less for Codex, and the reverse. OpenAI hasn’t said whether the five-hour window will cover chats.
• Pro. Pro plans “currently have no five-hour limit” in Work and Codex, and since October 29 Pro 100 and Pro 500 have twice their previous Work and Codex allowance, as our memo on it explains. The open question for Pro is GPT-6 Pro, which has its own weekly limit in Chat today and runs on Astra.
• Business. Named in the article. Standard Business has the same five-hour estimates as Plus, and workspaces with flexible pricing “can purchase additional workspace credits to continue working.”
• Free and Go. Not named. Both include limited Codex access, and everyday text chats are unlimited.
• Enterprise and Edu. Not named. Those with flexible pricing already have “no fixed rate limits,” says Codex’s pricing; their usage is paid for in credits.
• People who only chat with Instant. By OpenAI’s account, nothing changes.

How it came out

On October 29, Tibor Blaho, an engineer who regularly posts unreleased ChatGPT changes, posted a “new hidden shared usage limit checkout disclaimer” from ChatGPT’s web app: “Starting in December, plan usage will have a shared limit across all of ChatGPT and Codex.” A member of the Chinese forum LINUX DO found the same sentence in the checkout page’s code.

The day before, an X user asked how ChatGPT “unifying Chat and Work” would change usage limits. OpenAI’s Thibault Sottiaux replied: “Plan is to bring together best of both Chat / Work capabilities together at some point this year, but we’re going to proceed carefully here and take our time.” Blaho linked the two: “This is likely related to the ongoing work to combine Chat + Work.” OpenAI’s article now says as much.

The article is narrower than the checkout notice in two ways. It names three plans rather than “plan usage,” and it leaves Instant chats out of “all of ChatGPT.”

Anthropic went this way first. Its plans’ limits “are shared across Claude and Claude Code, meaning all activity in both tools counts against the same limits,” says its help center, and since September 16 it has been rolling out “everything Claude Cowork does” to every conversation on Pro and Max, by its release notes.

What OpenAI hasn’t said

• The day. “Starting in December” gives no date, and nothing says whether it waits for your renewal.
• The size. Whether the shared allowance is larger than today’s Work and Codex allowance, to make room for chats.
• How each kind of chat counts. Thinking, GPT-6 Pro, voice, images, file uploads and dots aren’t mentioned.
• The other plans. Free, Go, Enterprise and Edu aren’t in the article.
• The five-hour window. Whether Plus and Business chats count toward it as well as the weekly limit.

What to do now

1. Don’t change plans over it yet. Nothing changes before December. Plus is “$20/month (billed monthly),” and OpenAI does “not support annual billing” for it. A Pro cancellation waits for the end of the period: “You keep access until the end of the current billing period,” says the Pro tiers article. You can wait for the numbers.
2. Learn what a normal week uses. Settings → Usage in ChatGPT, or /status in the Codex CLI, shows your allowance and reset times. Knowing what a typical week takes lets you see the difference when chats start to count.
3. Use Instant for quick questions. Once the change lands, it’s the one kind of chat OpenAI says won’t count. Save higher effort and GPT-6 Pro for the questions that need them.
4. Keep a fallback for code. Plus and Pro subscribers who reach their limit “can purchase additional credits to continue working,” and “all users may also run extra local chats using an API key, with usage charged at standard API rates,” says Codex’s pricing. A second lab’s plan works too, such as Claude Code on a Claude plan, though Claude already shares one limit between chat and Claude Code.

When Codex runs out in the middle of a task

One limit or two, some weeks end early. Enjoy runs Codex on your ChatGPT plan, beside Claude Code, Antigravity, Grok Build, Muse Code and local models through OpenCode, and keeps the work going when a limit hits:

• Its usage meters show the limits Codex reports for each account, such as 5 hours and Weekly, with credits and free resets, before you start a big job.
• When Codex runs out, the conversation waits for the reset and carries on by itself, as long as the computer is awake and Enjoy is open. If you’ve signed in another ChatGPT account with usage left, it moves there instead.
• Or switch the conversation to another agent, such as Claude Code on a Claude plan or a free local model, and the new agent catches up on the work so far.

Enjoy doesn’t change OpenAI’s limits or add usage to a ChatGPT plan: each agent runs on the plan you already have. Enjoy is a monthly subscription, with the first seven days free (pricing).

We’ll update this memo when OpenAI names the day or publishes the new allowances.