OpenAI used DevDay, on 29 September 2026 at Fort Mason in San Francisco, to ship Dots: always-on agents inside ChatGPT, powered by GPT-6 Astra. Each Dot gets a cloud computer and a browser. The company described them as agents built to keep working after the person has closed the laptop.

That is a different object from a chat reply. A Dot is assigned a goal, can use a large plugin ecosystem, and is meant to continue in the background. OpenAI said a person can message or call their Dot, and can open the agent's computer to see what it is doing. Wired and CBS reported the same shape: a personal agent on a machine you do not own, depicted as a customizable figure, with specialist versions pitched for business work.

Who could use one

At launch, Dots began rolling out to ChatGPT Pro and Business Premium in eligible markets. Reporting at the time said Pro subscribers in the European Economic Area, Switzerland, and the UK were left out. One Dot was included with those plans. Tasks draw on plan allowances. Conversations with the agent, OpenAI said, do not.

The same keynote carried a cheaper model, GPT-6.1 Sol, described as near Astra on coding and office work at a lower price, and a higher Pro tier. The agent was the announcement that changed the product category. The model was how they proposed to pay for it.

The name that was not used

In the weeks before DevDay, reports described an unreleased model string that contained the word Aeon. OpenAI did not announce a product by that name. The public name was Dots. A rumor and a launch can share a month and still be different facts. This site will keep them apart.

Source: TechCrunch