OpenAI Dots are always-on AI agents announced on 29 September 2026 at OpenAI's DevDay. Powered by GPT-6 Astra, each Dot runs on its own cloud computer with a browser, connects to more than 4,000 apps through OpenAI's plugin ecosystem, and is designed to keep working toward goals between conversations.
This page is an independent reference guide. It is not OpenAI or ChatGPT. Product claims below prioritise OpenAI's own announcement and related documentation.
What a Dot is
OpenAI describes Dots as frontier intelligence that has the user's back: agents that learn preferences over time, take important work off the plate and stay available around the clock. Unlike a single-turn chatbot exchange, a Dot can be given ongoing responsibility, plan next steps, report progress and ask for decisions that need human judgment.
Users interact with a Dot through ChatGPT on desktop, web and mobile, and through Slack and Microsoft Teams. Voice calls are supported in ChatGPT. Text messaging was described as a limited beta for some US Pro users at launch. OpenAI has said that over time teams of Dots could work together on a user's behalf.
Architecture detail is on How Dots Work and in the post GPT-6 Astra and the Dot Cloud Computer.
Cloud computer, plugins and proactive mode
Each Dot has its own cloud computer and browser. Users can open that computer to inspect work. With permission, a Dot can also connect to the user's laptop. Through plugins it reaches a large app ecosystem—OpenAI cites more than 4,000 apps.
When the user is not actively chatting, a Dot may perform proactive research using read-only tools on connected apps. OpenAI states those background tools cannot send messages, edit app content or control the browser or computer. Assigned tasks can still use the broader toolset under the user's approval rules.
See Features and Permissions and Approvals.
Permissions and control
Users choose which apps a Dot may access and can set rules that allow actions, require approval or block them. Default safeguards plus custom instructions form the control model OpenAI emphasises. Conversations with a Dot do not count toward standard ChatGPT usage limits; tasks the Dot starts in Codex or ChatGPT Work follow those products' usual allowances.
More on controls: Privacy, Security and Safety.
Availability and pricing at launch
Dots rolled out to ChatGPT Pro and Business Premium. The first Dot is included at no extra cost. Enterprise, Edu and Healthcare workspaces can access a beta when a workspace admin enables it; the beta is off by default.
- Regions: Business Premium is described as available across supported ChatGPT regions. For Pro, eligible markets at launch excluded the European Economic Area, Switzerland and the United Kingdom.
- More Dots: OpenAI said users will later be able to add more Dots and scale a Dot's speed or monthly work volume. Per-Dot add-on pricing was not published on announcement day.
- Specialist Dots: OpenAI previewed organisation-scoped specialist agents with their own identity, IT-provisioned hardware options and deeper systems-of-record integrations for defined roles.
Dedicated coverage: Pricing and Plan Access, Who Gets Dots and Specialist Dots.
How Dots relate to other agents
Dots compete in the same always-on agent category as products such as Meta Muse. Differences include distribution (inside ChatGPT plans versus a consumer Muse app), model stack (GPT-6 Astra versus Muse Spark), plugin breadth claims and enterprise specialist positioning. Treat each product's primary documentation as authoritative for its own limits. A structured comparison is in Dots vs Meta Muse.
Primary source
- OpenAI: Introducing dots (29 September 2026)
This guide focuses on documented entities—GPT-6 Astra, cloud computer, plugins, Pro and Business Premium inclusion, regional limits, read-only proactive mode and specialist Dots—so readers can evaluate the product from primary materials. Method notes are on the About page.
Anya Ivanova
Original Post
No discussions yet. Be the first to start!