ChatGPT Dots: What They Are, Who Gets Them, Limits and Texting

Dots by plan

OpenAI publishes no fixed task count for a dot. Its plan includes "an allowance for deeper work", and OpenAI said dot usage would not count toward plan allowances for the first month after launch, with usage terms for each plan to follow. That month ran out around October 29, 2026; as of October 30 the release notes carried no usage terms, and the help center still described only the launch month's extended limits. OpenAI also plans to let you add more dots and scale each one by speed or by monthly workload, but has not announced prices. Access is rolling out gradually and can take several days to reach an account. Manus, by contrast, publishes credit costs for a few example tasks (Manus credits explained).

How usage is counted

1. Conversations with your dot do not count toward your ChatGPT usage limits.
2. Codex and ChatGPT Work tasks your dot starts or manages count toward your usage limits as usual, so a dot that launches coding tasks draws on your plan's Codex allowance, then credits.
3. The first month after launch, dot usage did not count toward eligible Pro, Business and Enterprise plan allowances. That window ran out around October 29, 2026, and OpenAI has not yet published the per-plan usage terms it promised for afterwards.

Budget a dot's Codex and Work tasks the same way you budget your own. ChatGPT Work and Codex limits by plan lists those allowances.

Dots in Codex and ChatGPT Work

A dot can use background agents, start new cloud threads and create Work or Codex tasks on a connected computer. It can continue existing local Codex tasks, check their results and send follow-up instructions to tasks it created. Cloud tasks need a Codex cloud environment you have already created in Codex, and local tasks need local computer access switched on in the ChatGPT desktop app. Work and Codex share one usage allowance on your plan, so a dot that runs coding work every day spends the same pool as your own Codex sessions.

What a dot does

A ChatGPT dot is an always-on agent powered by GPT-6 Astra and, since November 3, 2026, also GPT-6.1 Terra, drawing on different models for different parts of its work. It works on its own cloud computer and browser, connects to the apps you allow through ChatGPT's plugins, and keeps working toward goals between conversations. OpenAI's examples include watching customer feedback and preparing tested fixes as pull requests, revising launch materials when a product's scope changes, rerunning an analysis when new data arrives, and drafting show notes and social posts from an interview transcript for approval.

• Its own computer. You can open the dot's cloud computer at any time. Access to your own computer is optional and off until you allow it from the ChatGPT desktop app.
• Parallel projects. A dot can run several projects at once and can create Codex cloud tasks in environments you set up.
• Memory. It starts with your ChatGPT memory and learns how you work.
• Proactive research. When you are not working with it, a dot reads your connected apps with read-only tools that cannot send messages, change app content or control a browser or computer.

Where you use a dot

You can create your dot in the ChatGPT iOS or Android app, the desktop app (including on Windows) or ChatGPT on desktop web; mobile web is not supported. You can message or call it in the ChatGPT mobile app, and message it in Slack or Microsoft Teams. You can also text it if you are on Pro in the US (next section). At launch a dot cannot call you, and it cannot have its own standalone email address, although it can use a personal email account you connect. How dots compare with Claude Duet and Meta Muse on channels, plans and limits is in Claude Duet vs ChatGPT Dots vs Meta Muse.

Texting a dot is a US-only Pro beta over iMessage or RCS

Since October 28, 2026, OpenAI has been rolling out texting in beta to ChatGPT Pro users in the US. It is not available in Business or Enterprise workspaces, and access is limited, so not every Pro user has it yet. On an iPhone the dot uses iMessage, and on Android it uses RCS. To turn it on, open your dot's profile in ChatGPT, select Texting, verify your phone number and save the dot's contact card when it arrives. OpenAI's help center points to the desktop app for this step, while its docs also list the mobile app and desktop web.

Texting is one-to-one: only you can text your dot, and it texts only you, with text, photos and voice messages, using the same memory and plugins as in the app. The dot can text you first with updates, and when it needs approval it sends a link to review the request in ChatGPT. Group chats, phone calls and FaceTime are not supported over text. OpenAI sends and receives the texts through Linq, a third-party messaging provider, says this gives neither the dot nor OpenAI access to your other text conversations, and advises caution with sensitive information. Reply STOP to stop the dot's outgoing texts; message and data rates may apply.

Safeguards and control

Dots build on ChatGPT's existing protections with extra controls for the work they can do:

• Separate environment. Your own computer and its files stay separate unless you choose to connect them.
• Permissions and Custom Rules. You choose which apps a dot may access, and Custom Rules let you allow specific actions, require approval or block them. Built-in safety requirements always apply.
• Action review. Dots use auto-review to check actions that could affect your accounts or share information against your instructions, rules and safety requirements. Some sensitive tasks, such as changing a password, always stay with you.
• Credentials. For supported websites a dot can use saved passwords without exposing them to the model.
• Monitoring. Safety monitoring can pause or stop a dot's work, and Activity View shows progress, including background work.

On data use, OpenAI does not use content from Business, Enterprise or Edu workspaces to improve its models by default. On personal plans you control whether dots' conversations and work are used, and OpenAI says it does not train directly on proactive research or a dot's notes to itself.

Dots compared with other ChatGPT and Codex automation

A dot can start Codex or Work tasks itself, so the choice is not exclusive. The difference is who decides what happens next: with a dot, the agent plans the next step toward a goal; with the other options, you or your code define each run.

Specialist dots for organizations

Separately from personal dots, OpenAI is previewing specialist dots: agents with their own identity, credentials and access to company systems, set up to own well-defined responsibilities such as procurement, invoice processing or customer support. They start as focused enterprise pilots run with OpenAI's engineering teams, and OpenAI is working with Microsoft to manage them through Agent 365. They are not a self-serve plan feature.

Bottom line

A dot is worth trying if you are on Pro or Business Premium in a supported market and have ongoing work that spans several apps. Start with narrow permissions and approval rules, watch which tasks it launches in Codex and ChatGPT Work, since those draw on your normal allowance, and revisit your plan choice once OpenAI publishes the usage terms that replace the launch-month exemption. Outside ChatGPT, Meta's Muse is sold on its own weekly-allowance plans (Meta Muse usage limits by plan), Claude Duet comes with Claude Pro, Max and Ultra in the US, and Manus Cue is free in invite-only early access; AI personal agents by plan sets them side by side. On a plan without dots, ChatGPT Plus vs Pro covers what moving up to Pro adds.