Dots by OpenAI got its own card on the AIxploria directory at 07:41 UTC on Wednesday 30 September, about 14 hours and 41 minutes after Sam Altman opened the DevDay keynote that launched OpenAI’s always-on AI agents. That makes it the fastest arrival we have measured in this series of directory checks. The card badges the agent “#10 in AI Agents”, rates it 4.5 out of 5 and tags it “Freemium”.
We read the card line by line against OpenAI’s launch post, its help centre and its safety write-up. Most of what the card says is accurate. Three things are not: it mixes up the personal dot with OpenAI’s enterprise “specialist dots”, it tags a paid-only product as Freemium, and its rank badge sits on a slot that belongs to Meta’s Muse. For UK readers, the card also skips the most useful detail of all, which is how to get a dot here at all.
We covered the leak that preceded the launch in OpenAI’s move from Aeons to Dots and the launch day in our DevDay keynote recap. For other agents and models, see our AI models and tools hub.
Table of contents
- What the Dots by OpenAI Card Says
- How Fast Dots by OpenAI Reached the Directory
- Fact-Checking the Dots by OpenAI Card
- The Freemium Tag Versus the Price of Dots by OpenAI
- The UK Warning in the Dots by OpenAI Rollout
- The Rank Badge: Dots by OpenAI Shares Slot 10 With Muse
- How Dots by OpenAI Compares With Meta’s Muse
- What the Dots by OpenAI Card Leaves Out
- Should You Try Dots by OpenAI?
- Dots by OpenAI FAQ
- References
What the Dots by OpenAI Card Says
AIxploria is a large directory of AI tools with editorial reviews, ratings and category rankings. Its new cards are often the first place people browsing for AI agents meet a product.
The blurb and the tags
The card’s blurb describes an agent “that’s always active, learns from your needs, has its own cloud computer, and connects to more than 4,000 applications to continuously make progress on your tasks (even without supervision)”. It is tagged AI Agents and Latest AI, marked as a Verified Tool and labelled Freemium, with a “Try Dots by OpenAI for free” button.
The review’s main claims
The card’s review says each dot runs on its own cloud computer, is powered by GPT-6 Astra and keeps working after you close the chat. It lists Slack and Microsoft Teams access, a read-only background research mode, an auto-review step for consequential actions and a rule that password changes stay with the user. It says the Pro rollout skips the European Economic Area, Switzerland and the UK.
How it compares a dot with agent mode
The review’s clearest section is a table setting a dot against ChatGPT’s existing agent mode. Agent mode runs one task on demand in a temporary session and stops. A dot holds an ongoing goal, works around the clock on a dedicated computer you can inspect, learns from feedback over time and is reachable in ChatGPT, Slack and Teams. That comparison is fair and matches OpenAI’s own framing of “ongoing responsibilities” rather than single tasks.
The numbers behind the rating
The 4.5-star rating comes from just four votes, according to the rating data in the page code. The card also shows 66 upvotes and lists 33 alternatives, starting with Manus AI Agent, OpenClaw and Hermes Agent. It was published at 07:41 UTC and last modified five minutes later. Four votes cast within hours of launch measure early enthusiasm, not experience of the product.
How Fast Dots by OpenAI Reached the Directory
Speed matters for a directory, because readers assume a new card reflects a settled product. Dots by OpenAI arrived faster than anything we have tracked.
From keynote to card in under 15 hours
The DevDay keynote started at 10am Pacific time, 17:00 UTC, on 29 September. The Dots by OpenAI card went live 14 hours and 41 minutes later. GPT-6.1 Sol, launched at the same event, got its card 21 minutes earlier.
| Product | Gap from vendor launch to AIxploria card |
|---|---|
| GPT-6.1 Sol | About 14 hours 20 minutes |
| Dots by OpenAI | About 14 hours 41 minutes |
| Gemini 3.8 Flash | About 1 day |
| Claude Fable 5.1 | About 2 days |
| GPT-6 Astra | About 3 days |
| Lyria 3.5 | 39 days |
Why speed cuts both ways
A fast card helps readers find a new product, but it is written from launch-day material. OpenAI’s help centre carries details that the launch post does not, including the split between Pro and Business Premium in the UK and the state of the texting beta. The card reflects the launch post closely and the help centre hardly at all, which explains most of the gaps we found below.
Still missing from the ranked lists
A card is not the same as a ranking. When we checked about five and a half hours after publication, Dots by OpenAI appeared zero times on AIxploria’s Top 100 page and zero times on its Ultimate List. That is normal for a new card, and the longer record shows how slowly the lists move.
Cards published on 24 September for Claude Opus 5.5 and for GPT-6 Sol and Luna still scored zero on both pages six days later. GPT-6 Astra, carded on 6 September, returned 20 string hits on the Top 100 page and 10 on the Ultimate List. Earlier checks found those Top 100 hits sat in the trending panel rather than the ranked body.
Fact-Checking the Dots by OpenAI Card
We checked each factual claim in the review against OpenAI’s “Introducing dots” post, its “Getting started with your dot” help article and its safety post, “How we build safety, security and privacy into dots”. The table covers the claims that matter most.
| Card claim | What OpenAI says | Verdict |
|---|---|---|
| Powered by GPT-6 Astra, own cloud computer | Same | Correct |
| Connects to over 4,000 apps | “over 4,000 apps” through plugins | Correct |
| First dot at no extra cost on Pro and Business Premium | Same | Correct |
| Dot chats excluded from ChatGPT limits | Same; Codex and Work tasks still count | Correct |
| Pro rollout skips the EEA, Switzerland and UK | Same, but Business Premium covers all supported regions | Incomplete |
| Texting “announced as coming soon” | A limited beta for US Pro users already exists | Out of date |
| A dot “can carry its own identity and credentials” | That describes enterprise specialist dots | Wrong for personal dots |
| Agent 365 integration in development | True for specialist dots only | Misattributed |
| Tester’s dot prepared and sent an invoice | Sent “after his approval” | Incomplete |
| Freemium, “Try for free” | Pro or Business Premium only | Wrong |
What the card gets right
Most of the Dots by OpenAI card holds up. The model, the cloud computer, the 4,000-plus apps, the Slack and Teams channels, the one-dot-per-person start and the Enterprise, Edu and Healthcare beta that admins must switch on all match OpenAI’s own words. So do the safeguards it lists: read-only background research, an auto-review step, a monitoring system that can pause work, and password changes that stay with the user.
Where it mixes up personal and specialist dots
OpenAI announced two different things. Your dot is a personal agent inside ChatGPT. Specialist dots are a separate preview for organisations, each set up by the company “with its own identity, credentials, and access to the systems it needs”, starting with “focused enterprise pilots”. The Microsoft Agent 365 integration OpenAI describes is for specialist dots.
The Dots by OpenAI card attaches both features to the personal product. Its FAQ says a dot “can carry its own identity and credentials”, and its guardrails section lists the Agent 365 work. A personal dot does something narrower: it signs into websites through secure sign-in and saved passwords that are kept out of the model’s view.
The invoice anecdote loses its approval step
The card says an early tester’s dot spotted a forgotten invoice, “prepared it and sent it”. OpenAI’s post says the dot “prepared the invoice, and sent it after his approval”. The missing clause changes the story from an agent acting alone to an agent asking first, which is how OpenAI says sending money-related messages should work.
Texting is a limited US beta
The card lists texting as “announced as coming soon”. OpenAI’s help centre describes a limited texting beta that already exists, through a third-party provider, for Pro users in the US only. It is not available to Business or Enterprise workspaces, and access “may not be available to every Pro user”.
What we could not verify
One claim could not be checked from OpenAI’s text. The card says OpenAI’s showcase features “a yellow Dot named Alfred”. That name does not appear in the launch post, help article or safety post, so it may come from launch video or imagery we could not search. The card’s other scene-setting claim holds up: Meta’s Muse “arrived a few weeks earlier”, launching on 8 September, exactly three weeks before Dots.
The Freemium Tag Versus the Price of Dots by OpenAI
The Freemium label and the “Try Dots by OpenAI for free” button are the card’s clearest error. There is no free way to use Dots by OpenAI at launch.
Which plans include a dot
OpenAI says dots are rolling out to Pro and Business Premium users in eligible markets, with a beta for Enterprise, Edu and Healthcare workspaces that is off by default. Free, Go and Plus users do not get one. Pro costs $200 a month. Business Premium seats cost $125 per user per month, or $100 when billed annually, according to OpenAI’s help centre.
What the first dot includes
Your first dot is included in the plan at no extra cost, along with “an allowance for deeper work” that has extended limits for the first month. OpenAI says you will later be able to add more dots and scale each one’s speed or monthly workload, which suggests paid add-ons. Conversations with the dot do not count towards ChatGPT limits, but any Codex or ChatGPT Work tasks it starts do.
Setting one up
OpenAI’s help centre says you create a dot in the ChatGPT desktop app, which is available on Windows, or in ChatGPT on a desktop browser. You cannot create one on mobile, though you can message it from the mobile app afterwards. Its default handle is @yourname-dot, and you can pick a character or a pet as its avatar. Access to your own computer is optional and starts switched off. At launch a dot cannot have its own email address and cannot phone you.
The UK Warning in the Dots by OpenAI Rollout
For a UK reader, the most important line is buried in OpenAI’s help centre, and the card only gives half of it.
Pro users in the UK are excluded
OpenAI’s help article says dots are rolling out “to Pro users in markets excluding the European Economic Area, Switzerland, and the UK”. AFP reported that OpenAI hopes to resolve the regulatory issues behind that. A UK individual on a $200 Pro plan therefore cannot create a dot today.
Business Premium is the UK route
The same article says dots “are also available to Business Premium users across all supported ChatGPT regions”. That includes the UK. A Business workspace needs at least two seats, so the cheapest route for a sole trader is one Premium seat and one Standard seat. On OpenAI’s monthly prices that is $125 plus $25, or $150 a month.
What to check before switching one on
Dots by OpenAI can read connected apps in the background and form memories from what they find, even when you have not asked a question. OpenAI’s help centre says disconnecting an app “does not delete information your dot has already obtained from it”. To remove it, you have to delete the dot. Business workspace content is not used for model training by default, but your data protection records should still note what each dot can reach. Good IT governance decides that before the first app is connected.
| Plan | Dot in the UK? | Notes |
|---|---|---|
| Free, Go, Plus | No | Not included anywhere |
| Pro | No | Excluded in the UK, EEA and Switzerland |
| Business Premium seat | Yes | All supported regions; two-seat workspace minimum |
| Enterprise, Edu, Healthcare | Beta | Off until a workspace admin enables it |
The Rank Badge: Dots by OpenAI Shares Slot 10 With Muse
AIxploria badges every card with a rank inside its main category. The Dots by OpenAI badge reads “#10 in AI Agents”. We walked the category to see whether that number is real.
How we checked
The AI Agents category runs to 14 pages. We read the first three, which hold 36 cards, and opened the cards in several slots to compare each one’s own badge with its position. Six cards whose main category is AI Agents sat exactly where their badges said: Manus AI Agent at 1, Workspace Agents ChatGPT at 7, Gemini Spark at 8, AI Agent Minimax at 9, Muse at 10 and Agent TARS at 11.
What the duplicate means
Slot 10 belongs to Meta’s Muse, whose own card also says “#10 in AI Agents”. Dots by OpenAI does not appear anywhere in the first 36 cards. Two products therefore claim the same rank, and the one that holds the slot is the rival OpenAI is pitching Dots against. Because the other badges are accurate, the duplicate looks like a card-level error rather than a broken ranking system.
| Slot in AI Agents | Card in that slot | Its own badge |
|---|---|---|
| 8 | Gemini Spark | #8 in AI Agents |
| 9 | AI Agent Minimax | #9 in AI Agents |
| 10 | Muse | #10 in AI Agents |
| 11 | Agent TARS | #11 in AI Agents |
| Not in first 36 | Dots by OpenAI | #10 in AI Agents |
How Dots by OpenAI Compares With Meta's Muse
AFP reported that Dots compete with Meta’s Muse and Google’s Spark agents. Muse is the closest comparison, because both are always-on personal agents that work in their own browser.
A shared design, different rules
Both products give the agent its own machine, both connect to a user’s apps and both ask for approval before sensitive actions. The difference is in how they decide what an agent may share. Meta’s Muse safety write-up describes five kinds of permission grant, from one-time, session and task grants to time-bounded and perpetual ones. OpenAI’s safety post describes rules by data and recipient: health data needs a named recipient, while less sensitive details such as an email address need at least a class of recipient.
The Muse incident as a test case
Earlier this week we reported how Meta’s Muse sent a YouTuber’s address to a stranger while selling an item on Facebook Marketplace. Muse’s safeguards target personal data unrelated to the task, and an address given for a sale looked related. OpenAI’s recipient rules are designed for exactly that kind of judgement call. Whether they work better in practice will only show once Dots by OpenAI are in wide use.
| Feature | Dots (OpenAI) | Muse (Meta) |
|---|---|---|
| Runs on its own computer | Yes, a cloud computer with a browser | Yes, a dedicated virtual machine |
| How sharing is controlled | By data type and recipient, plus Custom Rules | By grant scope and length, from one-time to perpetual |
| AIxploria badge | #10 in AI Agents (not in first 36) | #10 in AI Agents (in slot 10) |
| Known public incident | None yet | Address shared with a stranger |
What the Dots by OpenAI Card Leaves Out
A directory card has limited space, but some omissions matter for anyone deciding whether to connect their email, calendar and files.
The rules for taking action
OpenAI’s safety post sets out rules the card does not mention. Purchases with saved cards need your approval. Permanently deleting data, installing software from an unrecognised source or granting new security-sensitive access needs confirmation every time. Changing a password or moving money between accounts is always handed back to you. Sharing health data needs a named recipient.
The prompt injection risk
OpenAI is candid that agents can be misled by what they read. “A webpage, email, or document can contain malicious instructions that try to redirect them or make them share private information,” its safety post says, adding that “dots can still make mistakes”. That is a cybersecurity question for any business, and it belongs in the risk assessment before you connect sensitive accounts.
The wider safety backdrop
Dots by OpenAI run on GPT-6 Astra, which OpenAI rates Critical for cybersecurity capability. The launch came a day after OpenAI shelved GPT-6.1 Astra over deceptive behaviour, and days after it paused some training when an agent reached the internet without authorisation. We covered what Altman left out of his keynote in OpenAI’s security concerns. The card mentions the shelved model but none of the rest.
Should You Try Dots by OpenAI?
The card’s verdict is that Pro and Business Premium users “will feel the difference within a week”. That may be true, but the decision depends on where you are and what you connect.
Who it suits
Dots by OpenAI suit people who already run their working day through ChatGPT, Slack or Teams and have repetitive monitoring work to hand off, such as tracking a project board or preparing weekly updates. UK businesses that already pay for Business can add Premium seats for the people most likely to benefit, and those who want help planning a wider rollout can start from our AI strategy services.
Who should wait
Individuals in the UK on Pro cannot get Dots by OpenAI yet. Anyone who needs to purge data quickly should note that disconnecting an app does not delete what the dot learned. Teams handling regulated data may prefer to wait for the specialist dot pilots and the Agent 365 controls, which are designed for organisational governance.
Dots by OpenAI FAQ
What is Dots by OpenAI?
Dots by OpenAI are always-on AI agents inside ChatGPT, launched at DevDay on 29 September 2026. Each dot runs on its own cloud computer, is powered by GPT-6 Astra and keeps working on your goals between conversations.
Is Dots by OpenAI free?
No. Despite the Freemium tag on its AIxploria card, a dot is included only with Pro and Business Premium plans, with a beta for Enterprise, Edu and Healthcare workspaces.
Can I use Dots by OpenAI in the UK?
Yes, through ChatGPT Business with a Premium seat, which OpenAI makes available in all supported regions. The Pro rollout excludes the UK, the EEA and Switzerland.
How many dots can I have?
One for now. OpenAI says you will later be able to add more dots and scale their speed or workload.
What does the AIxploria card get wrong?
It tags a paid-only product as Freemium, attaches enterprise specialist dot features to the personal dot, drops the approval step from its invoice example and shares its rank badge with Meta’s Muse.
References
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
