What are ChatGPT Dots?
ChatGPT Dots are personalized agents inside ChatGPT. You give a dot ongoing responsibilities, connect the apps it may use, and tell it what matters to you. It can then continue work in the background, keep context between conversations, and ask for input when a decision needs you.
OpenAI describes Dots as agents that can work on their own using a cloud computer, browser, and connected apps. You can inspect the computer and the dot’s activity, redirect its work, or pause it. OpenAI’s launch announcement also describes Dots working through ChatGPT, Slack, and Microsoft Teams.
The useful distinction is continuity. A standard chat usually responds to a request in the current conversation. A dot is designed to own a responsibility over time, pick up follow-up work, and bring back results for review.
Who makes ChatGPT Dots?
OpenAI makes ChatGPT Dots as part of ChatGPT. OpenAI’s launch announcement says the first dots are powered by GPT-6 Astra and notes that GPT-6.1 Terra has joined the family of models that can power them. OpenAI says a dot can use different models for different parts of its work.
The dot is the agent experience; the models are part of the system that powers its work. Its tools and access depend on the connected apps and permissions you provide.
What can a dot do?
A dot can handle ongoing work that spans multiple steps. OpenAI’s examples include tracking customer feedback, scoping small improvements, building and testing fixes, preparing pull requests for review, and attaching videos that show changes.
Other examples include updating analysis when new data arrives, keeping a proposal aligned with changing requirements, or preparing content from an interview transcript. These are examples of intended workflows, not a guarantee that every task will succeed without review.
For software teams, Dots may help carry bug work forward from feedback to a tested change. To make that practical, the agent needs useful evidence: what the user did, what went wrong, and the environment where it happened. See how clip.qa works with ChatGPT Dots for a bug-reporting workflow, or learn about AI-ready bug reports.
How do ChatGPT Dots work?
Dots use an agent-specific cloud computer, browser, and the apps you connect. They can keep working between conversations and may split larger work among background agents. You can review their activity and set rules for which actions are allowed, require approval, or are blocked.
OpenAI says proactive background research uses connected apps with read-only tools. Actions that could change accounts or share information go through review, and some require approval. Dots can still make mistakes, so consequential code, messages, and account changes should be checked by a person.
ChatGPT also supports custom remote MCP servers in supported plans and workspaces. OpenAI’s MCP documentation explains how to add a public HTTPS endpoint as a plugin and review the tools it exposes. Since Dots use connected tools, teams can make compatible MCP tools available through ChatGPT subject to account policy and permissions. See the ChatGPT MCP connection guide.
How do Dots compare with other agents?
Dots combine persistent work, ChatGPT context, connected tools, and Codex delegation. A coding agent focused on a repository may offer a more direct code-editing workflow; a conventional ChatGPT conversation is useful for a contained question but does not have the same ongoing responsibility model.
| Workflow | What it is suited to | How it differs from a dot |
|---|---|---|
| ChatGPT conversation | Questions and tasks handled in a conversation | A dot can carry an ongoing responsibility between conversations |
| Codex | Software engineering tasks in a coding environment | OpenAI says Dots can start and follow up on Codex work; Codex is one part of the workflow |
| ChatGPT Dot | Work that benefits from continuing context and connected apps | You set the responsibility, permissions, and review boundaries |
OpenAI’s examples show Dots coordinating work and returning fixes for review. They do not mean a dot replaces a developer’s judgment, source control, or test review. For another agent workflow, compare clip.qa with ChatGPT Dots with the report-first bug-fixing loop for AI-generated code.
How much do ChatGPT Dots cost?
OpenAI says the first dot is included at no extra cost with eligible Pro and Business Premium plans. Dots are rolling out in eligible markets. Enterprise users, including Education and Healthcare workspaces, can try the beta when an administrator enables access. OpenAI’s launch page says an allowance for deeper work is included, but it does not give a universal standalone price for additional dots.
Plan names and availability can change. Check the current ChatGPT pricing page and your workspace settings before choosing a plan. Some tasks a dot starts in Codex or ChatGPT Work count against those products’ usage limits, even though conversations with the dot do not count toward ChatGPT usage limits.
What are the limits of ChatGPT Dots?
Dots need app access, permissions, and enough context to do useful work. A cloud computer is separate from your local computer unless you choose to connect the local computer, so files and sessions on your device are not automatically available to a dot.
Availability is still rolling out, and workspace administrators control access in some plans. Custom MCP connections are governed by ChatGPT plan and workspace policies; they connect to remote servers, not directly to a local MCP process. OpenAI also warns that Dots can make mistakes, and that consequential work needs review.
For bug fixing, a vague description can still leave the agent guessing. A short screen recording can make the defect easier to investigate. clip.qa’s mobile apps and Chrome extension are built, but beta and store availability are not confirmed. The Chrome extension can capture browser console and network details; mobile reports can include device telemetry. No SDK is required in the app under test.
Want to give an agent clearer bug evidence? Sign up free to start with clip.qa.
Sources
- Introducing dots — OpenAI — capabilities, model family, availability, included dot, safeguards. Date checked: 2026-11-05.
- Dots — ChatGPT — product overview, connected tools, permissions, availability. Date checked: 2026-11-05.
- Create your dot in the ChatGPT mobile app — ChatGPT Learn — mobile setup and Codex handoff. Date checked: 2026-11-05.
- Connecting custom MCP servers in ChatGPT — OpenAI Help Center — ChatGPT custom remote MCP setup, limits, and plan requirements. Date checked: 2026-11-05.
- ChatGPT pricing — current plan names and pricing page. Date checked: 2026-11-05.