OpenAI announced Dots at DevDay on September 29, 2026. The company's own description is short: "Remarkably capable, always-on agents built to handle everything."

The part that matters most is in the same launch post. A dot "learns what you need and gets to work before you ask, quietly in the background, 24/7." Most AI assistants so far wait for a prompt. Dots are designed to keep working on a goal after you've set it.

This guide covers what Dots are, how they work, who can use them today, how OpenAI handles purchases and approvals, and what that design says about letting any agent near a trading account.

What Is OpenAI Dots?

A dot is a persistent AI agent inside ChatGPT, powered by OpenAI's GPT-6 Astra model. OpenAI's launch post pitches a wide range of work: a dot "can do simple things like book a table—or take on your most ambitious work with the initiative of a high-agency engineer or chief of staff."

Three things separate a dot from a normal ChatGPT chat:

  • It has its own computer. Each dot runs on its own cloud computer with a browser. According to MarkTechPost's launch coverage, you can open the dot's computer at any time to see what it's doing, and your personal devices stay separate unless you connect them.
  • It works across your apps. Dots can connect to 4,000+ apps "through ChatGPT, using the plugins you connect." You choose which apps to connect and manage their permissions. A dot can also use accounts you sign into in its browser, and connecting your own computer is optional.
  • It keeps running. A dot keeps working toward your goals in the background and learns from your feedback over time.

You set up your dot in the ChatGPT desktop app or a desktop browser. After that you can message it from the ChatGPT mobile app where available (mobile web isn't supported), and from Slack and Microsoft Teams. The launch announcement listed texting as coming soon, and OpenAI's help center describes it as a limited beta for some US Pro users. A dot can also start tasks in Codex and ChatGPT's work tools for research, documents and software.

How Dots Differ From ChatGPT and Codex

ChatGPT can already run scheduled tasks, and Codex can run background automations. The difference with Dots is ongoing responsibility: one agent that holds your goals across projects and apps.

Typical ChatGPT chat Codex Dots
Main job Conversation, plus scheduled tasks Software work, including background automations Ongoing goals across work and personal tasks
Runs where In ChatGPT In a coding environment On the dot's own cloud computer
Reach Apps connected to ChatGPT Code and repositories Apps you connect via plugins, accounts it signs into, and tasks it starts in Codex or ChatGPT
Initiative Mostly responds to you Mostly works on assigned tasks Designed to "get to work before you ask"

Who Can Get a Dot, and What It Costs

Dots are rolling out to paid plans in eligible markets:

  • ChatGPT Pro (plans from $100 a month) and Business Premium ($125 per user per month billed monthly, or $100 billed annually): the first dot is included at no extra cost. Deeper work runs on an allowance, and tasks a dot starts in Codex or ChatGPT count toward those products' usual limits.
  • Enterprise, Education and Healthcare: beta access, once a workspace admin enables it.
  • Free and Go plans: not included, per Implicator.ai's breakdown of the new plans.
  • Region: availability depends on market. At launch, Implicator.ai reported that Pro subscribers in the EEA, Switzerland and the UK couldn't use Dots yet.

For now you start with one "primary dot" that you name and set up. OpenAI says it envisions "teams of Dots working together on your behalf" over time.

OpenAI launched a new $500-a-month Pro 500 plan on the same day, with the fastest version of its top model and the highest Pro usage allowance. You don't need it to get a dot.

What Controls Does a Dot Have?

An agent that works before you ask needs limits. OpenAI's safety FAQ for Dots describes several layers:

  1. Read-only background work. When a dot researches proactively, it uses your connected apps through read-only tools. It can't send messages, change app content, or control your browser or computer during that work.
  2. Custom Rules. You can allow, block, or require approval for specific actions, such as reviewing a draft before a message goes out. Custom Rules can't remove OpenAI's mandatory confirmations, handoffs, or core safety requirements.
  3. Auto-review. Planned actions that could affect your accounts or share information are checked against your instructions, Custom Rules and safety requirements. Some go ahead, some need your approval, and some are handed back to you.
  4. Monitoring. OpenAI can pause or stop a dot if it detects a safety concern.

You can also inspect the dot's activity and open its computer to see what it's doing.

Can OpenAI Dots Spend Money?

Yes, in a limited way. According to OpenAI, a dot can make purchases using a card you've already saved on a merchant's website, and those purchases require your approval. Moving money between financial accounts is different: like changing a password, OpenAI's rules have the dot hand it back to you.

That's a sensible line. Routine errands (buy the thing, book the table) can be delegated with a checkpoint, while transfers stay with a person. OpenAI is also clear that these safeguards reduce mistakes rather than rule them out.

Can OpenAI Dots Trade Crypto?

The materials OpenAI has published don't describe crypto trading, wallets, or exchange integrations for Dots, and they don't announce any integration with LeverUp.

OpenAI's purchase rules are still a useful model for thinking about agents and trading. Dots get read-only access for proactive work, need approval to spend, and are required to hand transfers between financial accounts back to you. Trading doesn't fit neatly into either bucket. A leveraged order isn't a transfer, but it can lose money faster than most purchases.

Approving orders one at a time also has limits. Trading risk builds up across many orders: total exposure, leverage, and how fast losses accumulate. You can approve every single order and still end up with a position you'd never have agreed to as a whole. For a longer look at why a spending cap doesn't cover this, see A Budget Is Not a Risk Limit.

If you ever connect any always-on agent to a trading account, set up at least:

  1. A trade-only key. It can open and close positions but can't withdraw funds, and you can revoke it without the agent's help. AI Trading Agents Need Scoped Permissions explains why.
  2. Hard limits in code, not in the prompt. Maximum leverage, maximum total notional, allowed markets, and a daily loss stop. Asking the agent to "be careful" is not a limit.
  3. A decision layer between the agent and the venue. The agent proposes and deterministic code checks the proposal before anything executes. See The Decision Layer for AI Trading Agents.
  4. Logs you actually read. Start at small size and review every order the agent submits.

On LeverUp, the documented way to do item 1 is 1CT Hosted Agent mode. The trader's wallet authorizes a separate signing key to submit trade intents on its behalf. In the developer docs' words, authorization "grants permission to trade, not to move tokens," and it can be revoked onchain. Trades it places can still lose the collateral they use, and strategy-level limits like the ones in item 2 need to be built separately. How to Build a Grok Bot Trading Agent on LeverUp walks through the setup.

FAQ

What is OpenAI Dots?

Dots are always-on AI agents in ChatGPT, announced September 29, 2026 and powered by GPT-6 Astra. Each dot has its own cloud computer, can connect to 4,000+ apps through plugins, and keeps working toward your goals in the background.

How much does OpenAI Dots cost?

The first dot is included at no extra cost in ChatGPT Pro (from $100 a month) and Business Premium ($125 per user per month billed monthly). Enterprise, Education and Healthcare have beta access. Free and Go plans don't include Dots.

Can I have more than one dot?

At launch you start with one primary dot. OpenAI says it plans for teams of Dots working together later.

Is OpenAI Dots available in Europe or the UK?

Availability depends on market. At launch, Pro subscribers in the EEA, Switzerland and the UK were reported as not yet eligible, so check OpenAI's help center for current status.

Can a dot act without my approval?

Some of it. Proactive background work uses read-only tools. Actions that could affect your accounts go through auto-review, and depending on your Custom Rules and OpenAI's safety requirements they may need your approval or be handed back to you.

Can a dot buy things?

Yes. A dot can make purchases with a card you've saved on a merchant's website, with your approval. Transfers between financial accounts have to be done by you.

Can OpenAI Dots trade crypto for me?

OpenAI hasn't described crypto trading or wallet support for Dots. Even with an agent that could trade, approving orders one by one isn't a substitute for hard limits on leverage, exposure and losses.

Trade on LeverUp: app.leverup.xyz