OpenAI Dots Launch: Access, Usage and Controls
OpenAI Dots are GPT-6 Astra-powered agents with their own cloud computers, 4,000+ app connections, cross-channel access, and configurable approval controls.
OpenAI Dots are always-on agents powered by GPT-6 Astra, each with its own cloud computer and browser. OpenAI launched Dots on September 29 with connections to more than 4,000 apps, access through ChatGPT, Slack and Microsoft Teams, and controls for permissions and approvals. Rollout is starting for Pro and Business Premium users in eligible markets; Enterprise users, including Edu and Healthcare, can try the beta when a workspace administrator enables it. Read OpenAI’s Dots announcement.
The product change is persistence. A dot can keep working across projects, use connected apps and contact its user with progress or questions. OpenAI says it can learn preferences from feedback and carry context between supported channels. Dots are a product layer around GPT-6 Astra, not a new Astra model ID or API pricing change. The GPT-6 Astra model page covers the underlying model; this article covers the agent product and its operating boundaries.
What OpenAI Dots can access and where they work
Each dot gets a separate cloud computer and browser that its user can inspect. A dot can connect to additional devices, and a user may explicitly allow it to use a laptop. The cloud computer remains separate from the user’s computer and files unless that connection is granted.
OpenAI says its plugin ecosystem lets Dots connect to more than 4,000 apps. That number describes available integrations, not automatic access. Users choose which apps a dot can reach through existing ChatGPT app controls. A saved password can sign in to supported websites without exposing the password to the model.
ChatGPT is the main control surface across desktop, web and mobile. Users can message or call a dot there, while it can send progress updates, questions and decisions. Slack and Teams messaging are also supported, and texting is planned. Context is designed to follow the dot between supported channels.
OpenAI illustrates Dots with coding, launch planning, research, sales and content work. Those scenarios describe intended workflows; they are not published cross-customer benchmarks for accuracy, speed or business impact.
OpenAI Dots controls: read-only research and approvals
OpenAI separates background research from actions that change an account or external system. When a user is not actively working with a dot, it can perform proactive research through already connected apps. Those tools are restricted to read-only access: they cannot send messages, change app content, or control the user’s browser or computer.
When work could affect an account or share information, Dots use a different control path. Built-in rules determine when the agent may proceed and when it must ask. Custom Rules let a user allow, require approval for, or block specific actions. Auto-review checks proposed actions against the user’s instructions, Custom Rules and safety requirements. Certain sensitive actions, including changing a password, always remain with the user.
Activity View shows progress and lets the user redirect work. OpenAI says security monitoring can pause or stop a dot when it detects a concern. Users should still inspect consequential changes before a pull request is merged, a message is sent, a commitment is made, or connected data is modified.
For connected-app work, the site’s multiple-account plugin guide shows how to check the account and permission scope used by a task. Workspace owners can also compare Dots with recorded ChatGPT Enterprise admin controls.
Data use depends on the workspace and settings
OpenAI says content from ChatGPT Business, Enterprise and Edu workspaces is not used to improve its models by default. On personal ChatGPT plans, users control whether Dots conversations and work can be used for model improvement.
OpenAI also says it does not train directly on proactive research or a dot’s private notes to itself. Information from those processes may still be used when it helps inform an eligible conversation or task, depending on the user’s settings. Teams should check both workspace policy and personal data controls rather than treating the read-only research boundary as a complete data-use policy.
App permissions and model-improvement settings answer different questions: one controls what the dot can reach or do; the other governs whether eligible conversations or tasks may contribute to model improvement. Administrators should document both before enabling a beta for sensitive work.
OpenAI Dots access, first-dot allowance and usage limits
Dots are rolling out in ChatGPT to Pro and Business Premium users in eligible markets. The first dot is included in either plan at no extra cost. Enterprise users, including Edu and Healthcare customers, can try the beta when an administrator enables it. Rollout does not mean every eligible account receives access immediately, so the GPT-6 access guide helps separate plan eligibility from observed availability.
The included dot and its work allowance are different parts of the offer. A Pro or Business Premium plan includes the dot plus an allowance for deeper work, with extended limits during the first month after launch. OpenAI says users will later be able to add dots and scale their speed or monthly workload, but the launch article does not publish those future prices or quantities.
Conversation usage is separate from delegated task usage. Talking with a dot does not count toward ChatGPT usage limits. If the dot starts or manages a task in Codex or ChatGPT Work, that task counts toward the applicable limits as usual. Users estimating capacity should identify where the work runs instead of assuming every action through a dot is excluded.
Initial creation happens in the ChatGPT desktop app or a desktop browser. After setup, users can message the dot in the ChatGPT mobile app.
How the Dots launch differs from the DevDay recap
OpenAI’s DevDay 2026 recap is a conference-wide index covering models, Codex, APIs, plugins, workspace features and commercial announcements. The Dots launch is the primary source for this agent’s cloud computer, 4,000+ app connections, channels, read-only proactive research, approvals, data policy and allowance.
The intent is different: a DevDay recap helps readers scan the full announcement set, while this article answers Dots adoption questions—what the product is, who can start using it, what consumes an allowance, which permissions apply, and when a person must approve an action.