OpenAI Dots at DevDay 2026: The $100 Agent and Everything Else OpenAI Announced

OpenAI announced dots at DevDay 2026 on 29 September. A dot is an always-on agent inside ChatGPT that runs on GPT-6 Astra, gets its own cloud computer and keeps working while you are away. The cheapest way to get one is ChatGPT Pro at $100 a month. If you live in the EU, the UK or Switzerland, Pro will not get you one at all.

That price is the part most launch coverage left out. OpenAI dots arrive three weeks after Meta’s Muse, which is free to start, and seven weeks after Grok Bot, which comes with a $20 plan. Below: what a dot actually does and what it really costs on every plan. Then the one route that works in Europe and how dots stack up against Muse and Grok Bot. Last comes a recap of everything else OpenAI announced at DevDay 2026, from GPT-6.1 Sol to a $500 Pro plan.

The Key Takeaways

• What they are: OpenAI dots are always-on agents in ChatGPT, powered by GPT-6 Astra, each with its own cloud computer and browser.
• What they cost: no separate price. Your first dot comes with Pro 100, Pro 200 or Pro 500 ($100, $200 or $500 a month) or a Business Premium seat. Free, Go and Plus get nothing.
• The Europe catch: Pro users in the EEA, UK and Switzerland are excluded. Business Premium works worldwide, but a Business workspace needs two seats, so the floor there is about $120 a month.
• Usage: chatting with your dot does not count toward ChatGPT limits. Work it starts in Codex or ChatGPT Work does.
• Also at DevDay 2026: GPT-6.1 Sol at a fifth of Astra’s API price, a $500 Pro plan with Ultrafast speed, plugin extensions inside ChatGPT, Space for teams, and your ChatGPT plan usable in 16 partner apps.
• Against rivals: Meta’s Muse starts free and Grok Bot comes with a $20 Cursor Pro plan. A dot’s $100 entry price is five times Grok Bot’s.

What Are OpenAI Dots?

OpenAI dots are always-on AI agents that live inside ChatGPT and keep working between conversations. OpenAI pitches them as agents “built to handle everything.” Each dot runs on GPT-6 Astra, OpenAI’s flagship model, and has its own computer and browser in the cloud. You give a dot a responsibility, not a single prompt, and it follows through as things change.

OpenAI introduced dots on its own X account on launch day:

Introducing dots, powered by GPT-6 Astra.

Remarkably capable, always-on agents built to handle everything.

— OpenAI (@OpenAI) September 29, 2026

A computer of its own

A dot’s cloud computer is what makes it different from a normal ChatGPT chat. OpenAI’s documentation says the dot “can keep working even when your computer is off.” It can research, analyse data, prepare documents and build software there. You can open that computer at any time and press Take over to do a step yourself, then hand control back.

Your own Mac or PC stays separate unless you connect it. Local access is optional, off by default, and limited to one personal computer at a time. With it switched on, a dot can work with local files and start Work or Codex tasks on that machine.

Where you talk to a dot

You reach the same dot in ChatGPT on desktop, web and mobile, in Slack, or on a voice call from the phone button in its chat. Microsoft Teams works too, but OpenAI’s admin guide says Teams access is limited to an invite-only alpha for now. Switching channels does not reset its memory. A dot also carries notes about your preferences and ongoing work from one conversation to the next, on top of your normal ChatGPT memory.

Every dot has a name and a cartoon avatar. You choose its shape, colour, eyes, glasses and accessories. Its handle starts as @yourname-dot and changes when you name it.

What Can a Dot Do?

A dot handles ongoing work: it watches something, updates it as things change, and brings you the decisions. OpenAI’s own documentation gives five examples. A dot can keep a sales proposal current as seat counts change, or turn interview material into captions and show notes. It can keep launch slides in step with product changes and update a study as results arrive. For developers, it can investigate bug reports and prepare a fix for code review.

The most human example in the launch post is small. OpenAI says an early tester’s dot noticed he had forgotten to invoice a publication, prepared the invoice, and sent it after he approved it.

That is the whole pitch in one line: work done before you remember to ask.

Proactive research and background work

When you are not talking to it, a dot looks for ways to help. OpenAI calls this “proactive research.” It reads the apps you have connected, such as Gmail, Google Drive or GitHub, and suggests next steps. The tools it uses for this are read-only, so they cannot send messages, change content or control a browser.

For bigger jobs, a dot splits the work across background agents that run in parallel and report back. It can open separate ChatGPT threads, start Codex cloud coding tasks, and run recurring checks on a schedule you set. OpenAI says plugins connect a dot to more than 4,000 apps.

A first job to give your dot

Start a dot on one recurring responsibility with a clear “draft, don’t send” boundary. OpenAI’s own guidance is to name the sources, say what to track, say when to interrupt you, and ask the dot to confirm any schedule. Copy this and swap in your own inbox label and time zone:

Every weekday at 8 AM Central European Time, check my connected Gmail for messages labelled "Clients" from the last 24 hours. Make a short list of anything that needs a reply, a decision or a payment from me, with a one-line summary each. Draft suggested replies, but do not send anything. Message me in ChatGPT only if a deadline is in the next 48 hours. Confirm the schedule before you start.

Review the first two runs and correct what it gets wrong. A dot learns from that feedback, so a week of small corrections is worth more than a long first prompt.

How Much Do OpenAI Dots Cost?

OpenAI dots have no price of their own. Your first dot is included in an eligible plan “at no extra cost,” and the cheapest eligible plan is Pro 100 at $100 a month. Free, Go and Plus subscribers cannot create a dot. OpenAI’s launch post named the eligible plans without prices:

Dots will be available in ChatGPT on web, mobile, and desktop across Pro, Business Premium, and Enterprise users in eligible markets.

To get started, create your first dot in the ChatGPT desktop app or your desktop browser, connect your apps, and let it introduce itself.

— OpenAI (@OpenAI) September 29, 2026

“Pro” now means three plans. OpenAI’s Pro plans help page lists them, and the dots documentation says all three qualify:

Some launch-day coverage says the $100 tier is left out. OpenAI’s own pricing page says otherwise: its Pro card starts “From $100” and lists “Dot, your always-on agent.” Pro is billed monthly only. As of DevDay, OpenAI’s help page says Pro 200 is open to new subscribers again, but new Pro 200 accounts get a lower usage allowance than before. For a dot alone, Pro 100 is the plan to buy.

What counts against your limits

Talking to your dot is free of ChatGPT limits. OpenAI says conversations with your dot “don’t count toward your ChatGPT usage limits.” Tasks the dot starts in Codex or ChatGPT Work do count. They use your limits exactly as if you had started them yourself.

The first month is generous on purpose. OpenAI says plans include an allowance for deeper work, “with extended limits for the first month after launch.” After that, it will publish usage terms for each plan. Extra dots, more speed and more monthly work are planned as add-ons later, with no prices yet.

Can You Get OpenAI Dots in Europe?

Not on a Pro plan. OpenAI dots on Pro are limited to users “outside the European Economic Area, United Kingdom, and Switzerland.” OpenAI gives no date for when Pro users there will get dots. The exclusion covers every EU country plus Norway, Iceland, Liechtenstein, the UK and Switzerland.

The one working route in Europe is ChatGPT Business with a Premium seat, which OpenAI rolls out worldwide. A Business workspace needs at least two seats, so the cheapest valid setup is one Premium seat for you and one Standard seat for someone else.

That is a company workspace, not a personal account.

Your dot belongs to the Premium seat, and workspace rules apply: OpenAI does not train on Business content by default, and an admin controls the settings. For a freelancer or a two-person team already paying for ChatGPT Business, it is a reasonable path. For one person who just wants to try a dot, it is a lot of paperwork for an experiment.

Using a VPN to buy Pro from outside Europe is not a route we recommend. The help page ties eligibility to your market, and an agent that holds your email and passwords is the last account you want flagged.

How OpenAI Keeps a Dot Under Control

A dot checks risky actions before it takes them. OpenAI runs an automatic review on anything that could affect your accounts or share information. The review decides whether the dot can go ahead, needs your approval, or must hand the step to you. Some steps, such as changing a password, always stay with you.

You can add custom rules for specific actions. Each rule sets one of four behaviours: act without asking, act only when you explicitly asked for it, ask first, or hand the job to you. A sensible start is “ask first” for anything that sends a message or spends money. OpenAI stresses that rules are instructions the dot tries to follow, and that it “can make mistakes.”

Passwords get special handling. The dot signs in to websites through a private form, and it can use saved passwords without the model ever seeing them. On training, OpenAI says it does not train directly on proactive research or the dot’s private notes. On personal plans, your normal ChatGPT data controls apply.

Two controls work differently from what most people expect. Pause stops the dot’s main task, but delegated tasks and scheduled runs keep going until you stop them under Activity and Scheduled. And disconnecting an app does not delete what the dot already took from it. Only deleting the dot itself wipes its memory, conversations and schedules. OpenAI’s help center calls that button Reset; the newer documentation calls it Delete.

What OpenAI Dots Cannot Do Yet

OpenAI dots launched with six clear gaps, all listed in OpenAI’s own help pages and documentation:

OpenAI Dots vs Meta Muse vs Grok Bot

OpenAI dots, Meta’s Muse and SpaceXAI’s Grok Bot all do the same basic thing. Each is an agent with its own cloud computer that keeps working after you close the app. The differences are price, where you can get them, and what each is built for.

When a dot is the right pick

Choose OpenAI dots if you already live in ChatGPT, pay for Pro, and your work runs through Slack, Google Drive or GitHub. A dot can hand coding jobs to Codex and documents to ChatGPT Work, which suits software and document-heavy projects. It also runs on GPT-6 Astra, OpenAI’s flagship model. The catch is the $100 floor and the Europe exclusion.

When Meta Muse is the right pick

Choose Meta’s Muse for personal life admin in the US: shopping, bookings, bills and email. It is free up to a weekly limit that Mark Zuckerberg put at 100 million tokens, and it asks before it sends an email or buys something. Our full guide to the Meta Muse AI agent covers what it gets wrong.

When Grok Bot is the right pick

Choose Grok Bot if you want several agents for $20 and already use Cursor or SuperGrok. SpaceXAI lets you run many Bots at once and put them in a group chat to pass work between themselves. On individual Cursor plans, each tier includes its own weekly Grok Bot usage, rising from Pro to Pro+ to Ultra. Read what Grok Bot is and how it works, or the Grok Bot vs Claude Cowork comparison if Anthropic is also on your list. For the wider field, see the best AI agents right now.

If you do not need an agent at all

Many people reading about dots want something simpler. They want a quick way to ask a top model a question, summarise a PDF or draft a document on a Mac. That does not need a $100 plan or a cloud computer watching your inbox. Fello AI puts GPT, Claude, Gemini and Grok in one native app for Mac, iPhone and iPad, without connecting your accounts to anything.

Everything Else OpenAI Announced at DevDay 2026

Dots led a long list. OpenAI called DevDay 2026 its “biggest yet, with more than 20 major announcements across ChatGPT, Codex, our models, and entirely new forms of working with AI.” Three of them change something for most ChatGPT subscribers. The rest are in the table below, taken from the official DevDay 2026 recap.

GPT-6.1 Sol: near-Astra for a fifth of the price

GPT-6.1 Sol is an upgrade to GPT-6 Sol. OpenAI says it “nearly matches GPT‑6 Astra’s intelligence on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices.” In the API it costs $2 per million input tokens and $10 per million output tokens, the same as GPT-6 Sol, against $10 and $50 for GPT-6 Astra. Cached input drops to $0.10, half of GPT-6 Sol’s rate.

In ChatGPT, GPT-6.1 Sol is rolling out to Plus, Pro, Business, Enterprise and Edu in Codex and ChatGPT Work. It is not yet in Chat, and Free and Go are not included. Our guide to GPT-6 Sol and Luna explains that Work-versus-Chat split.

Pro 500 and Ultrafast

The new Pro 500 plan costs $500 a month and carries OpenAI’s highest usage allowance, “25 times the ChatGPT Plus allowance.” It is the only Pro plan with Ultrafast, a speed tier OpenAI says offers “up to 8× faster token generation (300 tokens per second) in Codex and up to 6x in the API.” GPT-6 Astra Ultrafast works in ChatGPT Work and Codex on Pro 500 and Enterprise, and in the API. It burns included usage at 8 times the Standard rate. Ultrafast for GPT-6.1 Sol is coming soon.

Pro 200 is open to new subscribers again, but at a lower usage allowance than before. Existing Pro 200 subscribers who qualify keep their old allowance through 29 October 2026, then move to the new one at the same $200 price.

Your ChatGPT plan in other apps

Sign in with ChatGPT now lets Plus and Pro subscribers spend their plan allowance inside other tools. OpenAI names 16 partners, including Cognition’s Devin, Notion, Vercel, T3, OpenClaw and Dactyl, and you can cap how much each one uses. Plan usage works in participating tools only. Signing in to other apps with your ChatGPT account works globally.

The full DevDay 2026 list

Every other DevDay 2026 announcement, with who gets it, according to OpenAI’s recap:

Should You Get an OpenAI Dot?

Get a dot if you are already on ChatGPT Pro outside Europe. It costs you nothing extra, chatting with it is outside your limits, and the first month has extended limits. There is no better time to find out whether an always-on agent fits your work.

Do not upgrade from Plus to Pro just for a dot. At $100 a month it costs five times the Cursor Pro plan that includes Grok Bot, and Muse starts free. OpenAI has not yet published what a dot’s allowance looks like after the first month. Wait for those terms. In Europe, only a Business Premium seat unlocks it, so the question becomes whether your team wants ChatGPT Business anyway. If you are weighing the plans themselves, our ChatGPT pricing guide breaks down what each one includes, and our GPT-6 Astra explainer covers the model every dot runs on.

FAQ