Best for professionals with a steady stream of recurring work, such as watching customer feedback, revising launch materials, or chasing invoices. It replaces manual check-ins rather than a chat window, and it improves as you correct it. Access is limited to two plan types in eligible markets, and Enterprise use is a beta that admins must switch on.
OpenAI dots is an always-on AI agent launched in ChatGPT on September 29, 2026. Each dot runs on GPT-6 Astra with its own cloud computer and browser, works around the clock, and connects to more than 4,000 apps through plugins. It reports through ChatGPT, Slack, and Teams, and asks for approval before consequential actions.
Maker: OpenAI · Autonomy: semi autonomous · Maturity: BETA
Underlying models: GPT-6 Astra
About OpenAI dots
OpenAI dots are always-on agents that live inside ChatGPT and work in the background on tasks you have handed them. OpenAI announced them on September 29, 2026, and describes each dot as an agent with its own cloud computer, its own browser, and the apps you have connected. That is what separates a dot from a chatbot: a chatbot waits for the next message, while a dot keeps several projects moving at once and messages you when it needs a decision.
Each dot runs on GPT-6 Astra and reaches more than 4,000 apps through OpenAI's plugin ecosystem. You can open the dot's computer at any time to inspect its work, and you can give it permission to connect to your laptop when you want it working alongside you. When you are not talking to it, the dot does what OpenAI calls proactive research, reading your connected apps with read-only tools so it can prepare work before you ask. It learns your preferences from feedback, so the tenth draft should need fewer corrections than the first.
The launch examples are concrete: a developer's dot watches customer feedback and returns finished pull requests with videos of the changes, a sales lead's dot builds a proof of concept for an integration, and a scientist's dot reruns analyses when new data lands. You reach a dot by text or voice call in ChatGPT on desktop, web, and mobile, and in Slack and Teams. Texting is listed as coming soon. If a task needs coding or document work, the dot can hand it to OpenAI Codex or ChatGPT Work, which are separate agents with their own usage limits.
Availability is narrower than the announcement's reach suggests. Dots are rolling out to Pro and Business Premium plans in eligible markets, and Enterprise, Edu, and Healthcare workspaces get a beta that an admin must switch on. Compared with self-hosted options such as OpenClaw or Hermes Agent, a dot trades control of the infrastructure for a managed, ready-made setup from OpenAI. OpenAI also previewed specialist dots with their own identity and credentials for enterprise pilots, and said it is working with Microsoft to manage them through Agent 365.
Pricing
OpenAI includes a first dot in ChatGPT Pro and ChatGPT Business Premium at no extra cost, with extended usage limits for the first month after launch. ChatGPT Pro is sold in three usage tiers and OpenAI's pricing page shows the Pro price as a starting figure that was not readable at review time, so no Pro price is stated here. Business Premium seats are reported at $125 per user per month, or $100 on annual billing (The Decoder). Enterprise-tier workspaces, including Edu and Healthcare, join through an admin-enabled beta at custom pricing. Conversations with a dot do not count toward ChatGPT usage limits, but tasks it starts in Codex or ChatGPT Work do. OpenAI says extra dots and higher speed or monthly output will be sold later.
Key Features
- Own cloud computer: Each dot works on its own cloud computer and browser, separate from your laptop, and you can open that computer to inspect what it did.
- Plugin reach: Connects to thousands of apps through plugins, so one dot can read a Slack thread and act in a project tracker.
- Proactive research: Between conversations the dot scans your connected apps with read-only tools and prepares work before you ask for it.
- Custom Rules and approvals: Custom Rules let you allow, require approval for, or block specific actions, and auto-review checks anything that could touch your accounts.
- Activity View: A running log shows what the dot did, including background work, so you can redirect it mid-task.
- Voice and chat everywhere: Reach one dot by chat or voice call across the ChatGPT apps, Slack, and Teams, and it keeps the same context in each.
Strengths
- Runs on its own cloud computer, so your laptop and its files stay separate unless you choose to connect them.
- Saved passwords can be used to sign in to supported websites without being exposed to the model, which limits a common agent risk.
- Your first dot is bundled into the plan, and chatting with it does not draw down ChatGPT usage limits.
- Sensitive actions such as changing a password always stay with you, whatever rules you set.
Weaknesses
- Available only on Pro and Business Premium in eligible markets, with Enterprise as an admin-enabled beta, so most ChatGPT users cannot try it yet.
- OpenAI states that dots can still make mistakes and tells users to review consequential work, so it is not a hands-off replacement for approval.
- Tasks a dot starts in Codex or ChatGPT Work count against your usage limits, so heavy delegation can still hit a cap.
- Texting is not live yet, and specialist dots for enterprises are limited to focused pilots.
Frequently Asked Questions
What does dots actually cost?
The first dot comes bundled with Pro and Business Premium and adds no separate charge, and OpenAI raised the usage allowance for the first month. The Decoder puts a Business Premium seat at about $125 a month per person, or $100 if you pay annually. Pro's own price was not readable on OpenAI's pricing page at review time, extra dots are promised for later, and Enterprise is quoted individually.
Are dots safe to connect to work accounts?
OpenAI says a dot is kept apart from your own computer unless you connect it, and that background research runs on read-only tools that cannot send messages or edit app content. Custom Rules let you block actions or require approval, and changing a password is always left to you. OpenAI's own advice is to review consequential work, since dots can still make mistakes.
What are OpenAI dots' closest competitors?
OpenClaw is the open-source choice if you want to host the agent yourself, and Hermes Agent is another self-hosted, model-agnostic option. Claude Code fits when the work is mostly software engineering in a terminal. None of the three ships bundled into a ChatGPT plan, which is the main convenience a dot adds.
How does dots compare to ChatGPT Work in 2026?
ChatGPT Work is a task-by-task agent you start, and its sessions count toward your usage limits. A dot is meant to keep running on standing goals and can start Work or Codex tasks for you, which then count against those same limits. If you only need one-off documents or spreadsheets done, Work is the simpler fit.
How do you set up dots?
Create your first dot in the ChatGPT desktop app or a desktop browser, name it, connect your apps, and let it introduce itself. After that first setup you can message it from the ChatGPT mobile app, Slack, or Teams. You need a Pro or Business Premium plan in an eligible market, or an Enterprise workspace where an admin has enabled the beta.
Top Alternatives
- ChatGPT Work: Pick ChatGPT Work for tasks you start and watch yourself in documents and spreadsheets; pick dots for standing work that continues without you.
- OpenAI Codex: Pick Codex when the whole job is a code change in a sandboxed repo; pick dots when the job spans several apps and channels.
- OpenClaw: Pick OpenClaw if you want an open-source agent you host yourself; pick dots if you want OpenAI to run the infrastructure.