ChatGPT Dots FAQ: 23 answers on cost, Europe, setup, three limit messages and on a break

The short version: what ChatGPT Dots is, who gets it, and what to check when it stalls

What it is

ChatGPT Dots (also called OpenAI Dots) is an always-on agent inside ChatGPT. You give it an ongoing responsibility and the sources it needs; it keeps working between conversations, brings results back, and asks when a decision needs you. It runs on GPT‑6 Astra and has its own cloud computer and browser, so work can continue while your own computer is off.

There is no separate Dots subscription. Your first Dot is included in ChatGPT Pro (Pro 100, 200 and 500) and in Business Premium. Free, Go, Plus and Business Standard were not part of the first rollout. Conversations with your Dot do not count toward ChatGPT usage limits; Work or Codex tasks it starts count toward those products as usual.

How to get started, in order

1. Open chatgpt.com/dots in a desktop browser, or use the ChatGPT desktop app or the latest mobile app. Mobile web is not supported.
2. Follow the introduction, name your Dot, and connect the one app your first job needs, or skip apps for now.
3. Give it one ongoing responsibility: what to keep track of, which sources to use, and which decisions stay with you.
4. Treat connections as separate. Messaging in Slack or Teams does not open your inbox; a plugin does not grant computer access; your computer connection is optional and off by default.
5. Review work under Activity in the Dot's profile. Recurring work sits under Scheduled. For the cloud computer, open Computers, then Take over, and Return control when you are done.

Where people get stuck, and what helped

• No Dots entry on an eligible plan. Access rolls out gradually and can take several days. Open chatgpt.com/dots on the right account and workspace, then wait or check how to get ChatGPT Dots.
• Personal Pro in Europe or the UK. Pro excludes the EEA, the UK and Switzerland, and OpenAI has not published a personal Pro date. Business Premium includes Dots in those regions. Details on EU and UK availability.
• "Your dot is on a break". That is an abuse-prevention pause on the Dot itself, not your plan running out. Timed reports lasted from about 1.5 hours to about 13.5 hours. Wait, open Activity for delegated tasks, and do not use Reset. Full playbook: "Your dot is on a break".
• "You've hit the preview limit. Try again at…". This stop names a time, and from mid-October many of those dates sat about a month out (13–15 November). Treat the date as a guide: some cloud-browser and Drive work ran again the same day in later reports. Note the time and timezone, then try one small action later.
• A delegated Codex or Work task says you are out of usage while Settings → Usage still shows room. cveralyon's Dot, Exion, had a Codex Cloud task rejected with a 14 November reset while Usage & limits showed 98% of the weekly agent allowance and 3,162 credits (openai/codex#54348). Open that task in Activity, screenshot both screens, and contact support if it lasts.
• You approved something and it is still blocked. Open the affected task for a separate sign-in, connection or approval. Asking for a draft does not authorize sending. dhruvashp's approved Gmail send was stopped by "Automatic approval review failed: You've hit the preview limit" before it ran; the mail was confirmed not sent while drafts and reads still worked (openai/codex#54306). Check whether the action went out before you retry.
• The Dot goes quiet with no break message. Open the task and answer any pending decision, approval, sign-in or connection there. If the Dot is paused, resume it. Each task has its own conversation, so correct it inside that task. Symptom map: Dots not showing or not working.
• Reboot vs Reset. Try Reboot first; it restarts the Dot and is meant to keep conversations, memories and scheduled tasks. Reset deletes the Dot. Several people who Reset to start over then saw "Failed to load your dot" instead of a fresh setup.

What isn't possible or known yet

• Personal Pro in the EEA, the UK and Switzerland has no published date.
• There is no public meter for the Dot's own abuse-prevention pause, and OpenAI has not published its threshold or reset length.
• At launch you get one Dot; more Dots and paid scale options are described as future, without prices yet.
• The Dot cannot initiate calls to you at launch. Texting is a limited US Pro beta, not for Business or Enterprise.
• Disconnecting an app does not erase what the Dot already read from it; only Reset deletes those notes, and Reset also deletes conversations and scheduled tasks.
• Ending a voice call or Pausing the Dot does not necessarily stop every delegated or scheduled task.

One starter message

Keep this as an ongoing responsibility, not a one-off chat. Use [name the sources: inbox, calendar, board, or repo]. Tell me when a deadline is at risk or a decision needs me; keep routine updates in a place I can open without asking you. Before any send, purchase or public post, show me a draft and wait for an explicit yes. Save progress as you go with the time of the last update at the top. If you hit "on a break", a preview limit or a usage limit, write down the exact message and any retry date, check whether a send or upload already went out, then tell me what is done and what isn't before you continue.

The basics

What is ChatGPT Dots?

It's an always-on AI agent inside ChatGPT, launched on 29 September 2026. You give your Dot an ongoing responsibility and the sources it needs, and it keeps working between conversations, messages you with results, and asks when a decision needs you. It runs on GPT‑6 Astra and has its own cloud computer and browser, so it keeps going while your own computer is off. More in What is ChatGPT Dots, and each capability, from the cloud computer and Codex tasks to calls and schedules, is explained on features.

How much does ChatGPT Dots cost?

There is no separate Dots subscription. Your first Dot is included in ChatGPT Pro (Pro 100, 200 and 500, at $100, $200 or $500 a month) and in Business Premium ($100 per user a month billed annually, or $125 billed monthly). Free, Go, Plus and Business Standard weren't part of the first rollout. The plan list is on pricing and usage limits.

Who can get it, and where?

Pro users aged 18 or over outside the EEA, the UK and Switzerland, Business Premium users in every supported ChatGPT region, and Enterprise, Edu and Healthcare workspaces once an admin switches on the beta. Access rolls out gradually, so an eligible account can wait a few days for the entry. See How to get ChatGPT Dots for the steps, and availability for who qualifies on each plan and in which regions.

Can I use ChatGPT Dots in Europe or the UK?

Not on a personal Pro plan yet, and OpenAI hasn't published a date. Business Premium does include Dots in the EEA, the UK and Switzerland. Details on ChatGPT Dots in the EU and UK.

Getting started

Where do I open it?

At chatgpt.com/dots in a desktop browser, in the ChatGPT desktop app on Mac or Windows, or in the latest ChatGPT mobile app. Mobile web doesn't work. Whichever device you start on, you get the same Dot everywhere.

How do I set it up?

Follow the introduction, connect the one app your first job needs (or skip it), name your Dot, then give it one ongoing responsibility: what to keep track of, which sources to use, and which decisions stay with you. Messaging, apps and your computer are separate connections. Getting started walks through the first week with real reports.

Does my computer have to stay on?

No. The Dot works on its own cloud computer. Connecting your own computer is optional and off by default; you only need it for local files, code or apps, and then that computer has to be online with the ChatGPT app open. You can connect one personal computer at a time. What people run on the cloud computer is on the cloud computer page.

Which apps can it connect to?

Gmail, Google Calendar, Google Drive, Notion, Slack, Microsoft Teams, GitHub and other ChatGPT plugins. Plugin permissions are shared with ChatGPT, Work and Codex, and you can allow reading without sending or writing. Setup and real cases for each app are on integrations.

Can I talk to it in Slack, Teams or by phone?

Yes. Add Slack or Teams as a contact method from the Dot's profile, or press the phone button in its conversation to call it. It is the same Dot in each place, with the same memory. At launch the Dot can't call you, and texting is a small beta for some US Pro users. See voice calls, Slack setup and the Dot as a Slack assistant.

Limits and problems

Does talking to my Dot use my ChatGPT limits?

No. Conversations with your Dot don't count toward ChatGPT usage limits. Work or Codex tasks the Dot starts or manages count toward those products' shared allowance, and deeper work has its own allowance, extended for the first month after launch. The split is explained on pricing and usage limits.

What does "Your dot is on a break" mean?

It's an abuse-prevention pause on the Dot itself, and people have hit it with plenty of plan allowance left. OpenAI hasn't published the threshold or when it lifts; in the reports where someone timed it, it lasted from about 1.5 hours to about 13.5 hours. A different stop, "You've hit the preview limit", names a retry time and from mid-October often pointed about a month ahead. A third stop, "You've hit your usage limit" on a Codex or Work task the Dot started, is that product's own allowance (cveralyon / Exion saw it while the Usage page still showed room). What to do for each is on "Your dot is on a break".

Why can't I see ChatGPT Dots, or why did it stop working?

If you never had it, check the plan, the region, the workspace and the rollout first. If it worked and then stopped, the fix depends on the symptom: a missing Dot, no replies, an unavailable cloud computer, calls that don't connect, or errors on your own computer. Dots not showing or not working sorts real reports by symptom.

My Dot stopped replying. What should I check?

Start with the work, not the chat. OpenAI's help article says to open the affected task and answer any pending decision, approval, sign-in or connection request there. If the Dot is paused, resume it; for recurring work, check the saved instructions and timing under Scheduled. Each task has its own conversation, so give corrections inside that task. If the screen says "Your dot is on a break" or names a preview limit, go to the on-a-break page; if nothing explains the silence, Dots not showing or not working collects reports of Dots that stopped replying.

I approved something, but it's still blocked. Why?

A yes in the chat doesn't always reach the step that needs it. OpenAI's help article says to open the affected task and look there for a separate sign-in, connection or approval request, then check the relevant custom rule: asking for a draft doesn't authorize sending it. Custom rules don't grant app or computer access or override built-in safeguards, and some steps, such as changing a password, only you can do. A stop can also be a preview limit mid-approval: dhruvashp's approved Gmail send failed before execution with a November retry date, and the mail was confirmed not sent (openai/codex#54306). For work handed to Codex, Dots steering Codex shows what finally got a yes through.

Should I use Reboot or Reset?

Try Reboot first: it restarts the Dot, and one user quotes its confirmation as saying conversations, memories and scheduled tasks are kept. Reset deletes the Dot with its conversations, memories and scheduled tasks, and several people who deleted a Dot to start over then saw "Failed to load your dot" instead of a fresh setup. Details in Dots not working.

Using it well

What do people actually use it for?

The jobs with the most first-hand reports on this site include steering Codex, building apps and sites, travel planning, cloud computer work, small business, voice calls and morning briefs. Every job page is on use cases.

How is a Dot different from Codex?

A Dot keeps a responsibility going across conversations and sources; Codex works on code in a repository. A Dot can start Codex or ChatGPT Work tasks itself and you'll find them under Activity in its profile. See Dots vs Codex vs ChatGPT Work and Dots steering Codex.

Will it send emails or change things without asking?

Built-in safety and approval checks are on from the start, and custom rules let you decide which actions it takes directly, which need your approval and which stay with you. Users still found loose wording risky: Alex Gopoian pointed out that "a draft to approve" can be read as permission to carry on. If you want to approve first, say that nothing is sent or changed until you reply yes. More on email triage.

Is it safe to leave a long job running overnight?

People do run overnight jobs with a Dot, but a pause can stop the job without any message in the chat. One user woke up to "on a break" and an unfinished Google Sheet. Ask the Dot to save progress as it goes, somewhere you can see it, and to tell you what's done when it's back. See project follow-up.

How do I pause or stop my Dot, or just one task?

These are three separate switches. Pause, in the ••• menu of the Dot's profile, stops only the Dot's current main task; select Paused • Tap to resume to start it again. Pause doesn't stop every delegated task or cancel future scheduled runs: open a delegated task in Activity to stop it, and disable or delete a recurring one under Scheduled. Stopping doesn't undo actions already completed, and ending a voice call doesn't necessarily stop work you assigned during it (Manage your dot).

Where do I find my Dot's work and files?

In the ChatGPT desktop app, select the Dot's name or avatar to open its profile, then open the task to see its progress, files and results. Tasks on your own computer appear in the sidebar on that computer, and recurring work is under Scheduled. Changing the Dot's selected computer doesn't move a task that already exists. To look at the Dot's cloud computer, open Computers in its profile, select Take over, and Return control when you're done. If tasks still seem to vanish, Dots steering Codex shows how people found them again.

If I disconnect an app, does my Dot forget what it read?

No. According to OpenAI's help article, disconnecting an app doesn't delete information the Dot already obtained from it. The only way to delete that information, and the Dot's own saved memories, is to delete the Dot with Reset, which also removes its conversations and scheduled tasks. So decide what an app should share before you connect it, and allow reading without sending where that is enough. App by app: integrations.

Can I have more than one Dot?

At launch you get one Dot. OpenAI says you will later be able to add more Dots and scale each one's speed or monthly workload, but it hasn't published prices for that yet.

Related

• What is ChatGPT Dots
• Getting started
• "Your dot is on a break": what it means and what to do
• How to get ChatGPT Dots
• Pricing and usage limits
• Dots not showing or not working
• Use cases by job