OpenAI's dots vs a scheduled Claude: one thing new.

// the verdict, and the sentence that decides it

One thing new, as of 2026-10-02. A dot has a browser in the cloud that keeps your website sign-ins and keeps working when your devices are off. OpenAI's docs:

Claude can browse too, but the browser lives on your computer. Anthropic's help page for Claude in the cloud, which already describes the October 6 change:

So a Claude task that needs a website with no connector waits on your computer. A dot doesn't. That is the whole gap. OpenAI adds one caveat of its own: "Some websites block cloud browsers or require extra verification."

What isn't new: picking its own wake-up time

The headline feature in OpenAI's docs is timing:

We expected that to be the second thing a scheduled Claude can't do. It isn't, quite. Claude Code, which comes with the same $20 Pro plan, has a self-paced /loop. Anthropic's docs:

The honest difference is scope. Claude's loop lives in a session ("Tasks only fire while Claude Code is running and idle") and expires after seven days. A dot is one standing agent that keeps its own notes across conversations, Slack, Teams and calls. That is a nicer way to work, but it isn't a capability a scheduled Claude lacks outright.

If you just use the app: its own example is a schedule

OpenAI's docs give this example under "A scheduled check-in":

Its own Tasks page adds: "Work that repeats at fixed times needs a saved schedule." That is a scheduled prompt, and Claude has one on every paid plan. Anthropic: "Scheduled tasks are available on all paid plans (Pro, Max, Team, Enterprise)", on a cadence of "hourly, daily, weekly, on weekdays, or manually", and they "can search Slack, query files, run web research, generate reports" through your connectors. After October 6, "Scheduled tasks run with no device online."

• Price: dots come with ChatGPT Pro ("Plans at $100, $200, or $500 USD per month", per OpenAI's pricing docs), Business Premium and Enterprise. Claude Pro is "$20 if billed monthly".
• Where: OpenAI lists Pro for "users over 18 outside the European Economic Area, United Kingdom, and Switzerland", and "Dots are rolling out gradually."
• Email: we don't claim a scheduled Claude can send email for you. Anthropic's own Gmail pages disagree on that.

If you write code: routines take a call from anything

Claude Code routines run in Anthropic's cloud with three kinds of trigger:

Routines are on Pro and up, are in research preview, have a one-hour minimum interval, and "Claude can use every tool from an included connector, including writes, without asking for permission during a run."

Dots' docs describe triggers too: a time, or "When a connected service supports event monitoring" an event such as a new bug report in Slack. We found no per-dot endpoint for your own scripts in those docs. Actions that affect your accounts go through an automatic review that decides whether the dot "can proceed, needs your approval, or must hand a step over to you", and custom rules can allow "Take action without asking".

What would change this verdict

If Claude's browser moves to the cloud, so it can use a signed-in site with your computer off, the one thing is gone and dots become a scheduled prompt with a nicer front end, at five times the entry price. If OpenAI's docs drop the cloud browser or its saved sessions, same result. Check Anthropic's "Use Claude Cowork on web, desktop, and mobile" page for the line about the desktop app.

Two more lines from OpenAI's docs worth knowing before you judge the price. On limits: "Conversations with your dot don't count toward your ChatGPT usage limits. Tasks your dot starts or manages in Work or Codex count toward those products' usage limits as usual. Your plan includes an allowance for deeper work, with extended limits for the first month after launch." On availability: "Dots are rolling out gradually. You may not see dots immediately, even if your plan is eligible." If the usage allowance shrinks after the first month, the value side of this comparison moves even if the capability side doesn't.

Going the other way: if your work lives in sites with no connector (vendor portals, booking sites, internal dashboards) and has to happen while you're away, that browser is worth paying for. If it lives in Gmail, Drive, Slack or GitHub on a schedule, a $20 Claude already does it. More on what makes something an agent: agents.

Sources: OpenAI, "Meet dots" (learn.chatgpt.com/docs/dots), "Tasks and memory", "Connect computers and apps to your dot" and "Control your dot" (learn.chatgpt.com/codex/dots/…), and "Pricing" (learn.chatgpt.com/codex/pricing). Anthropic, "Schedule recurring tasks in Claude Cowork" (support.claude.com 13854387), "Use Claude Cowork on web, desktop, and mobile" (15520349), "Use the built-in browser in Claude Cowork" (16607400), "Claude Cowork and chat are one Claude" (16761823), claude.com/pricing, and Claude Code docs "Run prompts on a schedule" (code.claude.com/docs/en/scheduled-tasks) and "Automate work with routines" (code.claude.com/docs/en/web-scheduled-tasks). openai.com and help.openai.com could not be read from our build environment, so nothing here relies on them. All checked 2026-10-02. OpenAI, ChatGPT, Claude and Anthropic are trademarks of their owners; this page is not affiliated with them.