Tools
OpenAI's agents that keep working without you come only to paid plans
OpenAI unveiled Dots on September 29: assistants that keep working on your tasks when you log off. For now they are for paid plans, such as Pro, from $100 a month; Muse, from Meta, has a free version.
The essentials
1 confirmed fact · 4 according to the source · 3 open questions
- OpenAI unveiled Dots on September 29, 2026, at its DevDay conference: agents that keep working on your tasks even if you log off.134
- According to the sourceAccording to WIRED, they are coming to ChatGPT's Pro plan, from $100 a month; MarkTechPost adds Business Premium. Each user starts with one Dot.345
- According to the sourceAccording to MarkTechPost, each Dot runs on its own computer in the cloud and can use more than 4,000 apps that the user connects.478
- According to the sourceAccording to WIRED, Dots ask for permission before sensitive actions, such as installing software, and users can set their own limits for them.34
- According to the sourceMuse, Meta's rival agent, has a free version; MarkTechPost also mentions paid plans and says it is available only in the United States.457
Why it matters
A chatbot waits for you to ask. A Dot takes the initiative: it looks at your calendar, messages you without being asked and keeps working on the task while you do something else. If it works as OpenAI promises, the assistant stops being a window where you ask questions and becomes someone you delegate to. The price of that convenience is trust: to be useful, it needs access to your accounts.
It is also a commercial fight. WIRED and The Verge read it as OpenAI's answer to Muse, Meta's agent, which has a free version356. Altman said a version for the general public can be expected, without giving a date5.
If you don't pay for one of those ChatGPT plans, nothing changes today. If you try it, think carefully about which accounts you connect to it: to be useful, it needs access to things like your email, your calendar or your files.
The details
According to MarkTechPost, you give a Dot a goal, a name and some limits, and it works from its own computer in the cloud, separate from yours, which you can open to check what it is doing4. You message it through ChatGPT, Slack or Microsoft Teams, WIRED reports3. In an OpenAI demo, a Dot saw in the calendar that the user would be working through dinner and suggested two delivery orders, with their prices3.
When you haven't given it a task, the Dot looks for ways to help on its own. In that mode it can only read the connected apps: it doesn't send messages or change anything, according to MarkTechPost4. Users can set rules on which actions they allow, which they block and which need their approval34. OpenAI admits that Dots can make mistakes and that work with consequences needs to be reviewed, the same outlet adds4.
The sources disagree on who can use it. WIRED mentions only the Pro plan35. MarkTechPost adds the Business Premium plan, in supported countries, and a trial version for companies, schools and healthcare organizations if their administrator turns it on4. According to this outlet, the first Dot is included in the subscription at no extra cost4. Sam Altman, OpenAI's chief executive, said it is starting as a high-end product because it uses a lot of computing power5.
For businesses, Altman announced Dots specialized in tasks such as accounting, email marketing or legal analysis3. The WIRED journalist who tried one says it works, but is keeping their guard up before letting it automate their life5. MarkTechPost notes that, a day before the launch, OpenAI disclosed that its agents had published images of 53 ChatGPT users on the internet4.
What we don't know
- When it will reach cheaper or free plans: Altman took it for granted, but gave no date.
- Which countries it is available in. MarkTechPost mentions “supported markets” without saying which, and no source mentions Europe.
- How it performs outside demos: the sources consulted include no independent measurement, and the performance figures they cite come from OpenAI itself.
Background
Dots run on GPT-6 Astra, OpenAI's model3. At the same DevDay, the company unveiled GPT-6.1 Sol, a model cheaper than Astra, according to the newsletters TLDR AI and Latent Space78. Muse, the Meta agent that Dots is being compared with, came out on September 8, 2026, according to MarkTechPost4.
Sources
- 1Introducing dots
- 2DevDay 2026 Recap
- 3OpenAI’s Dots Are Always-On AI Agents—and Its Answer to Meta’s Muse
- 4OpenAI Launches dots: Always-On GPT-6 Astra Agents That Work From Their Own Cloud Computers
- 5The Battle to Be Your Personal AI Agent Is Here
- 6OpenAI’s new agent is a shot at Meta — but can it compete with free?
- 7OpenAI Dots 🟡, GPT-6.1 Sol ⚡, software factories 🏭
- 8[AINews] OpenAI DevDay 2026: Dots, 6.1 Sol, Ultrafast, Decisions API, Agents API, Spaces, Marketplace, and 1.2 Billion ChatGPT WAU
- ·OpenAI’s Dot agent is enterprise software that can also order your dinner
- ·OpenAI connects the dots on always-on agents
- ·OpenAI DevDay 2026
Was this story useful? Yes Not really
This story was written in Spanish by an artificial intelligence system from the sources listed above, and translated automatically. Before publication, a program checks that every figure and every name appears in the sources, and that the translation keeps the same facts, figures and sources. It has not been reviewed by a person. Spanish original · How we work (in Spanish)