OpenAI Launches dots: Always-On GPT-6 Astra Agents That Work From Their Own Cloud Computers
MarkTechPost Michal Sutter ● Covered by 22 sources
OpenAI launched dots: persistent GPT-6 Astra agents with their own cloud computer. They keep working after you log off, but only inside OpenAI’s managed setup.
Based on reporting by MarkTechPost, Michal Sutter — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI used DevDay to introduce dots, a new kind of ChatGPT agent that is meant to stay on the job. Each dot gets its own cloud computer and browser, plus a single identity and a standing assignment instead of the usual back-and-forth prompt cycle. The pitch is simple: give it a goal, set boundaries, and let it keep going.
That setup is less radical than it sounds. OpenAI says dots can reach 4,000+ apps through ChatGPT plugins, and they can also kick off work in Codex and ChatGPT Work for research, analysis, documents and software. They can be reached in ChatGPT on the web, desktop and mobile, and Slack and Teams are supported too, with texting on the way.
The more interesting part is what happens when nobody is watching. In background mode, connected apps are read-only, so the dot can look for ways to help without sending messages or changing content. OpenAI also says you can open the dot’s cloud computer whenever you want to inspect what it’s doing, which is the kind of transparency these systems need if they are going to be trusted with real work.
Under the hood, OpenAI is betting on GPT-6 Astra. In its latency simulations, the model scored 72.6% on OSWorld 2.0 at roughly 40 minutes per task, ahead of GPT-5.6 Sol at 65.7% and roughly 75 minutes. That is a meaningful gap, though OpenAI is also clear that dots still make mistakes and that consequential work needs review.
So the company wrapped the agent in a lot of control machinery. Custom Rules can allow, block or require approval for certain actions. Auto-review handles anything that might affect accounts or share information. Activity View shows progress, and monitoring can pause or stop a dot if something looks wrong. Sensitive tasks like changing a password stay with the user, even though saved passwords can be used for sign-in.
Dots are rolling out now for Pro and Business Premium users in eligible markets, with Enterprise, Edu and Healthcare workspaces getting a beta once an admin turns it on. The first dot is included at no extra cost in Pro and Business Premium, and OpenAI says conversations with a dot do not count toward ChatGPT usage limits. There is no self-hosted or open-weights option.
My take — AI-written commentary, not fact-checked reporting
This is OpenAI doing what OpenAI does best: packaging an old idea until it looks inevitable. Persistent agents are useful, but the real product here is control — and the company knows nobody wants a cloud gremlin freelancing inside their email. The closed-system approach makes sense, even if it leaves the open-model crowd muttering into their terminal.
Read more about this at: MarkTechPost
