Picture this. It is seven in the morning, and you have not typed a single prompt yet. But your research is already done, flights compared, and a draft reply to your client waits in your inbox. Nobody stayed up all night to do it. One of OpenAI’s ChatGPT Dots did it while you slept.
That is the promise behind ChatGPT Dots, OpenAI’s always-on personal agents. Until now, only the priciest plans had them. On November 19, OpenAI began testing them with a small group of ChatGPT Plus subscribers on the $20-a-month plan. According to Engadget, it is the biggest step yet toward an assistant that needs no asking.
What Are ChatGPT Dots, Exactly?
Dots are not another chatbot window. OpenAI introduced them at its DevDay conference on September 29 as proactive agents that handle ongoing tasks across your connected apps. You give one a goal, and it keeps working on it in the background, even when you close the tab or put your phone down.
Each Dot gets its own cloud computer and its own web browser. That is the detail that makes them different from a normal ChatGPT session. A regular chat stops when you stop talking. A Dot keeps going, checking sources, filling in forms, and making progress on a project while you are offline.
Through OpenAI’s plugin system, Dots can reach more than 4,000 apps. That reach is what turns them from a clever demo into something genuinely useful. They can pull information from your calendar, your email, and your documents, then act on it across the services you already use.
The ChatGPT Plus Test: Who Gets In
The test was announced by Thibault “Tibo” Sottiaux, the engineering lead for Codex and ChatGPT, in a post on X. “Heads up, we’re starting to test dots with a small number of ChatGPT Plus users,” he wrote, adding that recent improvements had produced encouraging feedback on how proactive the agents have become.
The key word is “small.” Paying $20 a month does not guarantee access. A Dot may simply appear in your account without warning, and OpenAI’s support team cannot switch the feature on for you. The company has not said how many Plus accounts are in the test, or whether the rollout will become permanent.
There are limits on who can take part. The minimum age is 18, and the test excludes the European Economic Area, Switzerland, and the United Kingdom. OpenAI has given no timeline for opening Dots to every Plus subscriber, and it has said nothing about bringing them to the free tier.
Business users are getting a clearer deal. On the same day, OpenAI began rolling Dots out to all ChatGPT Business Standard users, its entry-level plan for companies. That is a full launch, not a test, and it suggests OpenAI sees workplaces as the first place where always-on agents will prove their worth.
Dots Get New Tricks on iPhone
The Plus test arrived with new features, several aimed at iPhone owners. Dots now have iOS widgets, so you can glance at what your agent is doing without opening the app. They also integrate with Apple’s Shortcuts automation, letting a Dot trigger, or be triggered by, the routines on your phone.
The most telling addition is messaging. Dots now support iMessage and RCS, the modern text standard on Android phones, and you can reply to a Dot straight from an iOS notification. The agent is moving out of the ChatGPT app and into the places where you already spend your time.
Two Brains Now Power ChatGPT Dots
When Dots launched in September, they ran on a single model: GPT-6 Astra, OpenAI’s flagship at the time. Earlier this month, OpenAI added a second option, GPT-6.1 Terra, giving each Dot a choice of brains depending on the job at hand.
The pairing reflects how OpenAI now builds its models. Astra is the heavy lifter, built for the hardest reasoning and research work. Terra is the lighter, faster option for everyday tasks. A Dot can pick the right one for the job, rather than burning flagship compute on a simple errand.
The timing is no accident. OpenAI has been shipping new models quickly this month, including GPT-6.2 Astra, which we covered in our report on the company’s new drug-discovery benchmark. Dots are the delivery vehicle for that intelligence, and every upgrade makes the agents in the background sharper.
Why OpenAI Is Opening the Door Now
Dots started life as a premium perk. At launch, they went to Pro subscribers paying $200 a month, plus Business Premium and Enterprise customers, with one free Dot included for Pro and Business Premium users. That kept them out of reach for the vast majority of ChatGPT’s paying audience.
The Plus test changes the maths. Plus is OpenAI’s mass-market plan, and its subscribers are the users most likely to benefit from an agent that handles the boring work. If Dots prove themselves with this group, they become a reason to stay subscribed, and a reason for free users to upgrade.
There is competitive pressure too. The race to build agents that act for you is now the hottest contest in AI. Perplexity has just brought its Computer agent to Chrome and other Chromium browsers through a free extension, while Google has shipped a Gemini agent for Mac that can manage files and fill in forms.
OpenAI cannot afford to keep its own agent locked behind its priciest tier while rivals hand theirs to everyone. The Business Standard rollout points the same way. Companies are where always-on agents can show measurable value fastest, in hours saved and tasks completed.
The Catch: Always-On Means Always Watching
An agent that works while you sleep needs access to do its job. It reads your email, checks your calendar, and signs into services on your behalf. That is the bargain at the heart of Dots, and it is one users should think about before they celebrate.
OpenAI says Dots learn from feedback over time, improving as they watch how you work. The more they know, the more useful they become. But the same knowledge that lets a Dot book your flight is the knowledge that makes a leak or a mix-up far more damaging than a stray chatbot reply.
Rival agents have already run into this problem. Meta’s Muse agent was recently reported to be mixing up conversations and private data across accounts, a reminder that always-on access carries always-on risk. OpenAI has announced no new privacy controls alongside the Plus test, so trial users are trusting the company’s existing safeguards.
The practical advice is simple. If you get access, start small. Give your Dot low-stakes jobs first, watch what it does, and check its work before you hand it anything sensitive. An agent earns trust the same way a new employee does: one completed task at a time.
What Happens Next for ChatGPT Dots
OpenAI is watching the test closely. Sottiaux said the company is looking forward to continued feedback as it expands, which suggests the Plus trial will grow if the early results hold up. A wider rollout could follow in the coming weeks, though OpenAI has committed to nothing.
The timing also fits a wider push. OpenAI has just rolled out ChatGPT Space, a shared workspace for Plus subscribers, in the same week.
A redesign is also coming. OpenAI is giving the Dot its own tab in the ChatGPT mobile app, starting with Pro users who already have one. Plus subscribers in the test will not necessarily get it, which underlines that this is still an experiment, not a finished product.
For now, the most interesting question is what people actually do with an agent that never sleeps. The first wave of Plus testers will answer it in practice, one delegated task at a time. If they come back with stories of mornings saved and chores quietly finished, Dots will stop being a premium curiosity.
The assistant that works while you sleep is no longer a demo for the $200-a-month crowd. It is knocking on the door of the $20 plan. The only question left is whether you will be one of the first to let it in.



