This article contains no product affiliate links.
What happened
OpenAI announced dots at DevDay on 29 September 2026 in the United States, early on 30 September in Japan. The company describes them as always-on agents: AI that can work towards a goal, rather than only answer individual questions.
The announcement says dots are powered by GPT-6 Astra and each has a cloud computer and browser. They can connect through a plugin ecosystem covering more than 4,000 apps. OpenAI calls background, read-only investigation of connected information “proactive research”. Dots can be reached through ChatGPT, Slack and Teams.
The help centre describes a gradual rollout to Pro users outside the European Economic Area, Switzerland and the UK, and Business Premium users in supported ChatGPT regions. Japan is not in the stated Pro exclusion list. Enterprise workspaces, including Edu and Healthcare, can try a beta when an administrator enables it. Access can take several days, and dots are not available to users under 18.
A first dot is included in Pro or Business Premium without an additional charge. Conversations with a dot do not count towards ChatGPT usage limits, but tasks it starts or manages in Codex or ChatGPT Work count as usual. The Pro tiers listed are US$100, US$200 and US$500 per month. Initial setup uses the desktop app or desktop web; mobile access follows where available. Text messaging is a limited beta for US Pro users.
Before the announcement, BleepingComputer reported on 27 September that a Pro benefits listing briefly included “o”, described as an always-on assistant. This is a report about a pre-release listing, not independent verification by this site. The sources reviewed for the Japanese article do not confirm that “o” and dots are the same product. The official DevDay recap includes more than 20 announcements, including GPT-6.1 Sol, Pro 500 and ChatGPT Space.
Why it matters
- The announcement moves beyond a chat that waits for a question towards AI that can work on ongoing responsibilities.
- Earlier reports about an always-on assistant created interest in its eventual name and availability.
- More capable agents make access permissions, human review and privacy important parts of the product.
An editorial note
OpenAI says dots can still make mistakes and consequential work should be reviewed. Its controls keep certain sensitive actions, such as password changes or money transfers, with the user; auto-review checks relevant actions, and Custom Rules can set boundaries such as not sending email. On personal plans, model-improvement settings control eligible conversations and work. Start with only the connections you need and tasks that involve reading information. Check current regional availability and plan conditions before paying for Pro.
Sources and dates
Original article: 2026-09-30. English edition prepared: 2026-10-01. Key official sources rechecked on that date. Details can change; consult the linked official pages.
- OpenAI: Introducing dots (29 September 2026) ↗
- OpenAI: DevDay 2026 Recap (29 September 2026) ↗
- OpenAI Help: Getting started with your dot ↗
- OpenAI Help: Dots privacy, security, and safety FAQs ↗
- OpenAI: How we build safety, security, and privacy into dots ↗
- OpenAI Help: About ChatGPT Pro tiers ↗
- Pre-announcement reporting: BleepingComputer (27 September 2026) ↗
