OpenAI Dots: Always-On AI Agents That Work Around the Clock

Imagine hiring a digital assistant who never sleeps, keeps a cloud computer of its own, learns from your feedback on every task, and keeps pushing toward your goals while you are offline. That is precisely what OpenAI launched on September 29, 2026 at DevDay: Dots, always-on agents that work 24/7 and are now included for every Pro subscriber.

These are not chat windows you open, use, and close. A dot runs continuously on OpenAI's infrastructure. You brief it once, and it executes across days. The mental model is closer to a dedicated remote employee than to a conversational tool — a change in the product's very shape, not another model upgrade.

Source: OpenAI official YouTube channel

What exactly is a dot? The official definition, verbatim

OpenAI's official email to Pro subscribers describes the product precisely: "remarkably capable, always-on agents built to handle everything. Powered by GPT-6 Astra, dots have their own cloud computer, learn from feedback over time, and can work towards your goals 24/7."

Unpack that sentence and you get four technical facts:

1. The engine: GPT-6 Astra, OpenAI's strongest and most expensive model — a dot is not a watered-down build.
2. The architecture: every dot gets its own cloud computer, so it can perform real work rather than just generate text.
3. The memory: it learns cumulatively from your corrections, improving with use instead of starting from scratch.
4. The persistence: it works toward your goals continuously, and conversations with your dot — and work it does directly — are included in your subscription without drawing on your ChatGPT usage allowance.

That last point matters more than it looks. Every direct interaction with a dot is excluded from your usage meter, which is an implicit admission from OpenAI that token-metered billing no longer fits agents that run around the clock.

Source: OpenAI official YouTube channel

Quick comparison: a dot versus a regular ChatGPT conversation

What this means for you: the pricing plans changed completely

The announcement did not arrive alone — it restructured the entire Pro tier:

• The new Pro 500 tier: the highest usage allowance OpenAI offers — 25× the Plus allowance — including Ultrafast with up to 8× faster token generation. Ultrafast is the line OpenAI built with Cerebras, which we covered in depth when it launched at 750 tokens per second on GPT-5.6 Sol; it is now an official feature inside the top subscription.
• Pro 200 changes from October 30, 2026: the included ChatGPT Work and Codex allowance drops from 20× to 10× the Plus allowance, and the GPT-6 Pro chat limit falls from 200 to 100 messages per week. The price stays the same.
• Compensation for current subscribers: a one-time grant of 62,500 usage credits worth $2,500, expiring December 31, 2026.
• A public commitment: the five-hour limit will not return, so you can spend your weekly allowance whenever you like.

For a professional building a product or running a distributed team from Manila to São Paulo, the equation is clear. If you use agents seriously, a dot on Pro 500 offers effectively unmetered operation on the strongest model available. The launch rode alongside the new GPT-6.1 Sol model Pro subscribers now use daily. If you are a moderate user, check the quota reductions before your next renewal — especially given the backdrop of OpenAI pausing new Pro sign-ups under Astra demand, pressure that visibly still shapes the company's capacity decisions.

How to start with your first dot

The practical steps are straightforward, per the official help article titled "Getting started with your dot":

1. Confirm your Pro subscription is active — dots are available to all Pro subscribers.
2. Open the dots section in the ChatGPT interface and create a new dot.
3. State its goal in one specific, measurable sentence; vague goals produce vague work.
4. Hand over its tools: the files and permissions it needs on its cloud computer.
5. Review its output regularly and give feedback — those corrections are the fuel for its cumulative learning.

The golden rule: start with a small, measurable objective before you delegate anything critical. Always-on agents need management like any new hire; the difference is that their mistakes bill at a far lower rate.

Limitations to weigh before the enthusiasm

First, dots are exclusive to Pro subscribers for now — no Plus, no free tier, and no announced expansion date. Second, "doesn't draw on your allowance" covers direct conversations and work; OpenAI has not documented every edge case, so treat heavy first-month usage with caution. Third, a dot runs on Astra: extremely capable, but it still errs like any autonomous system, and reviewing its output remains your responsibility — doubly so for anything touching clients or sensitive data. Fourth, dots availability follows Pro availability in your country, so verify payment and tier access in your region before building plans around it.

Finally, a point many miss: enabling a dot on a goal means granting it execution autonomy. Start with reversible objectives and raise the autonomy ceiling gradually as trust builds — and expect a weekly review of your dot's work to become part of your routine, much like checking your inbox.

Before you enable your first dot: a checklist

Before pressing the create button, run through this quickly. Is your subscription Pro? Is your goal written as one specific, measurable sentence? Have you defined what "done" looks like? Are the required files ready to share? Have you set a weekly review slot for feedback? Five yes answers mean you are genuinely ready — not just excited.

And if your studies or professional work need an assistant that works with you today rather than next quarter, platforms like Truescho already ship smart study tools — flashcards, quizzes, and summaries — while the culture of always-on agents matures around you.

Frequently asked questions

What are OpenAI Dots?

Always-on AI agents launched by OpenAI on September 29, 2026. Each dot runs around the clock powered by GPT-6 Astra, owns a dedicated cloud computer, learns cumulatively from your feedback, and performs real work toward long-term goals instead of only answering prompts.

How much do OpenAI Dots cost?

Dots are included in the Pro subscription at no separate price, and conversations with your dot plus work it does directly do not count against your ChatGPT usage allowance. For heavier use, OpenAI announced the new Pro 500 tier with 25× the Plus allowance and Ultrafast generation up to 8× faster.

Are dots available to Plus or free users?

No. The official announcement makes dots available to all Pro subscribers only. OpenAI has not announced a date for expanding them to Plus or the free tier. History suggests wider rollout eventually, but there are no public commitments.

What is the difference between a dot and GPT-6.1 Sol?

GPT-6.1 Sol is a language model you call on demand through OpenAI's products or API. A dot is an always-on agent running on GPT-6 Astra with its own cloud computer and cumulative learning — the difference between a car engine and a driver you employ by the day.

What is the new Pro 500 subscription?

A higher subscription tier announced alongside dots on September 29, 2026. It provides OpenAI's highest usage allowance (25× the Plus allowance) and includes Ultrafast mode with token generation up to 8× faster, aimed at professionals running agents intensively.

Sources

• Official dots announcement on OpenAI's site — the official launch page
• OpenAI Help — Getting started with your dot — the official getting-started guide
• Official Introducing dots video — OpenAI's official YouTube channel
• OpenAI DevDay 2026 keynote — OpenAI's official YouTube channel