ChatGPT "dots" are OpenAI's new always-on AI agents. Announced at DevDay on 29 September 2026 and powered by the GPT-6 Astra model, a dot is an assistant you hand a standing responsibility — not a single prompt. It has its own cloud computer and browser, keeps working between conversations, and messages you with findings and decisions to approve.
The demos are genuinely eye-catching: a dot that drafts your social posts over the weekend, catches a pricing mismatch across launch files, or has your next piece of content ready before you finish the current one. The question for a commercial team is not whether that looks magical. It is whether it saves time once you account for briefing, checking, correcting, and the subscription.
This guide strips the launch noise back to what is verifiable: what a dot actually is, who can get one (more restrictive than first impressions suggest — and it excludes individual EEA, UK, and Swiss accounts), what it really costs, the claims worth questioning, and the handful of ecommerce and marketing workflows where it earns its keep today.
It is written for ecommerce operators, DTC and Amazon brand teams, and agency marketers deciding whether to put real money and attention behind this now. The one-line verdict: promising for continuity work a capable person already knows how to do, not a replacement for that person's judgment.
OpenAI now has four ways to work inside ChatGPT, and the names blur together. The cleanest way to tell them apart is by what you hand over.
| Way of working | You hand over | Best for |
|---|---|---|
| Chat | A question | Brainstorming, quick research, a single draft |
| ChatGPT Work | A finished outcome from your files | A campaign brief, a report, a deck, a cleaned export |
| Codex | A software outcome | Building a tool, a landing page, a small app |
| A dot | An ongoing responsibility | Keeping a launch moving, watching competitors, queuing the next brief |
The first three wait for you to ask. A dot does not. You give it a goal, point it at the right sources, and it keeps working in the background, bringing results back and asking for decisions. The practical shift is from "what do you answer when I write" to "what do you do while I'm not looking."
Mechanically, each dot runs on its own cloud computer and browser, so it can keep going after you close the tab. It can reach you inside ChatGPT (desktop, mobile, and web), in Slack, and in Microsoft Teams, and you can call it by voice; text messaging is listed as coming soon. It connects to more than 4,000 apps through plugins, and it can delegate the actual production work to ChatGPT Work or Codex rather than doing everything itself.
This is where the hype tends to get loose. Dots are not on the plan most people pay for, the regional fine print rules out a lot of readers, and the "$200 or $500" framing is wrong. Here is the verified picture as of 5 October 2026, from OpenAI's own Help Center.
| Plan | Monthly price (USD) | First dot included? | Notes |
|---|---|---|---|
| Free / Go / Plus | $0 / $8 / $20 | No | Not eligible for a dot |
| Business Standard | $25 / seat ($20 billed annually) | No | Not eligible for a dot |
| Pro 100 | $100 | Yes | The cheapest way to get a dot; ~5× Plus usage |
| Pro 200 | $200 | Yes | More usage; new-subscriber allowance cut from 30 Oct 2026, same price |
| Pro 500 | $500 | Yes | Adds Astra Ultrafast and the highest usage allowance |
| Business Premium | $125 / seat ($100 billed annually) | Yes | Available in all supported regions |
| Enterprise / Edu / Healthcare | Custom | Beta, off by default | A workspace admin must switch dots on |
The $100 entry point. Your first dot is included at no extra cost in every Pro tier and in Business Premium. Nothing about the agent is reserved for the $500 plan. If you are choosing a Pro tier purely for dots, Pro 100 gets you the same dot as Pro 500 — what the higher tiers add is more usage and, on Pro 500 only, the faster Ultrafast speed mode. The common claim that a dot requires the $200 plan — with $500 as the premium tier — does not match OpenAI's pricing page.
The regional catch — read this if you are in Europe or the UK. Individual Pro plans (100, 200, and 500) are rolling out in markets that exclude the EEA, Switzerland, and the United Kingdom. So a Germany-, EU-, or UK-based person on a personal Pro subscription does not currently get a dot, at any tier. Do not upgrade a personal Pro plan to "unlock dots" from one of these regions — you won't get them. The one route that is available here is Business Premium, which is rolling out in all supported regions: a UK or EU company on Business Premium seats can use dots even though a UK or EU individual on Pro cannot. Enterprise accounts can also enable the beta.
What "always-on" costs in usage. Talking to your dot does not count toward your ChatGPT usage limits. But any task it runs in ChatGPT Work or Codex does count against those allowances — so a busy dot still draws down your plan. You currently get exactly one dot; OpenAI says it will later sell additional dots and let you pay to scale a dot's speed or workload, but has not published those prices. Treat "always-on" as a scheduling convenience, not unlimited labour for a flat fee. Also note Pro is billed monthly only, with no annual discount, and Pro 200's included allowance for new subscribers drops on 30 October 2026 while the price stays at $200.