Quick answer
OpenAI Dots are ongoing agents powered by GPT-6 Astra that can research, follow scheduled tasks, and keep working between conversations. Creation starts on desktop, while access is rolling out by plan and region. Before connecting services, review permissions, memory behavior, and action rules; OpenAI says Dots can make mistakes and consequential work should be checked.
Dots are for ongoing responsibilities, not just one-off prompts
OpenAI describes a Dot as an always-on agent that can take responsibility for a project, use its own cloud computer, and make progress between conversations. Dots run on GPT-6 Astra and can work with apps you choose to connect. You can review progress, change direction, or pause the agent. OpenAI also describes scheduled reminders and background research; its help center says proactive research has restrictions and cannot directly send messages, change content through plugins, or control a browser or computer.
Sources: OpenAI, OpenAI Help Center, OpenAI Help Center
Access depends on plan, region, and rollout
OpenAI says Dots are rolling out gradually. Pro access excludes the European Economic Area, Switzerland, and the United Kingdom at launch; Business Premium access is listed across supported ChatGPT regions. Enterprise, Edu, and Healthcare workspaces can try a beta when an administrator enables it, and it is off by default. OpenAI says the first Dot is included in Pro and Business Premium plans at no added cost. Check your account because rollout can take several days. Create a Dot in the desktop app or desktop web; mobile is for access after setup, not creation.
Sources: OpenAI Help Center
The included plan allowance is not a promise of unlimited work
Pro and Business Premium plans include an allowance for deeper work, with extended limits during the first month after launch. OpenAI says it will share each plan's usage terms after that period. The announcement does not establish unlimited agent activity. Treat the allowance and rollout as plan-specific, and check the current product terms before assigning a recurring or lengthy task.
Sources: OpenAI Help Center
Review connected data and actions before handing over work
Connected-app permissions are shared with other ChatGPT products, and a Dot may review permitted sources proactively and retain information in its context. Disconnecting an app stops new access but does not erase information already incorporated. Local-computer access starts off and requires an explicit connection. Custom Rules can shape some actions, but OpenAI says they cannot override built-in safeguards; Dots can still make mistakes, so review important work. For personal plans, the model-improvement setting governs whether eligible conversations and work may be used to improve models; Business, Enterprise, and Edu data are not used for training by default.
Sources: OpenAI Help Center, OpenAI Help Center
A background agent is a different tool from a multi-model chat app
Dots are designed to keep a task moving between conversations. A conversational multi-model app is better evaluated on the models and tools it actually exposes for a question or document workflow. These are different jobs: compare access, connected-data permissions, review points, and limits rather than assuming a chat model picker provides an always-on agent. Chat AI's public model directory shows its current catalog; it does not establish access to OpenAI Dots.
Frequently asked questions
What readers usually ask
Can I create a Dot on my phone?
Not at launch. OpenAI says to create one in the desktop app or desktop web. Mobile access is available after setup as the rollout reaches your account.
Are Dots available to every ChatGPT user?
No. Access depends on plan and region, and rollout is gradual. Pro access initially excludes the EEA, Switzerland, and the UK; Business Premium and enterprise workspace access have separate conditions.
Does disconnecting an app erase information a Dot already used?
No. OpenAI says disconnecting stops new access but does not delete information already incorporated into the Dot's context. Review its memory and deletion controls before connecting sensitive sources.
Will a Dot always ask before taking action?
Not necessarily. Custom Rules can allow some actions without asking, but built-in safeguards still apply, and some sensitive actions require approval or handoff. OpenAI also warns that Dots can make mistakes.
Evidence