ChatGPT Dots vs Grok Bot vs Tasks in Gemini: Which Is Worth Paying For?

Before upgrading to ChatGPT Pro for a Dot, check two alternatives: Grok Bot and Tasks in Gemini. Both belong in the same buying decision, and your current subscription may already include the ongoing help you need.

Start with your existing access. Eligible Google AI Pro users should consider Tasks in Gemini before adding another agent subscription. Eligible ChatGPT Pro users should consider their included Dot. If you are buying specifically for ongoing help across applications, Grok Bot deserves a place on the shortlist alongside both.

For reminders, a daily digest or a repeatable briefing, ordinary scheduled tasks may already cover the job.

This comparison examines the agent capabilities within ChatGPT, Grok and Google Gemini. It is based on current product documentation, checked October 19, 2026. We have not run a matched performance test, so the recommendations concern access, cost and workflow fit rather than speed or accuracy.

First, clear up Google's name change

Google now calls Gemini Spark Tasks in Gemini. Its task guide describes a Chat/Task toggle beside the prompt box and says existing tasks and schedules carried over. The Privacy Hub also uses the new name.

Three similar names belong to different things. The Google Tasks to-do app is a list the agent can update. Scheduled actions in ordinary Gemini chat are a separate, simpler feature. The business Gemini agent Google Cloud introduced on October 8 is a business coworker for Workspace companies, which is a different product from the personal feature compared here. Some help pages and older guides still say Spark, so check their access details against the current requirements below.

What does the eligible route cost?

These are standard US dollar prices before applicable taxes, using direct subscription routes. Local prices, billing channels and availability can differ. An included agent still has usage limits.

Sources: Google subscriptions, X Subscription pricing, Grok Bot plan access, ChatGPT Pro tiers and Business seat billing.

For a ChatGPT Plus subscriber paying $20/month, moving to the lowest Pro tier adds $80/month. Adding a $30 Grok Bot route while keeping Plus adds $30 instead. The work allowance and your actual usage determine how far either payment goes.

The Business minimum also matters. With mixed seat types permitted, one Premium seat plus one Standard seat totals $150/month, or $1,440/year with annual billing. That annual commitment is equivalent to $120/month, and only the Premium user gets Dots. These totals are our arithmetic from the published seat prices, not a separately advertised bundle.

Older guides quote $20 a month for Grok Bot through a Cursor plan. Existing Cursor and SuperGrok subscribers can keep those plans, but new self-serve customers start on the X Subscription, where Grok Bot begins at Premium. The $8 Plus tier does not include it.

Eligibility can decide the choice before features do

Tasks in Gemini requires an adult's personal Google account, Google AI Pro or Ultra, Keep Activity enabled and Personal Intelligence opt-in. Google's eligibility guidance excludes the EEA, Nigeria, Switzerland and the UK. Ordinary managed work and school accounts are not eligible for this personal feature.

A small-business owner using managed Google Workspace needs an eligible business route. Paying for personal Google AI Pro will not unlock this feature on the managed account.

Dots eligibility also depends on plan, region and rollout. Dots requires users to be at least 18. Personal Pro access excludes the EEA, Switzerland and the UK. Business Premium has a broader supported-country route. Enterprise, Edu and Healthcare access is a beta controlled by administrators, rather than an automatic entitlement for every employee.

For small teams, Grok Premium seats start at $30 per user monthly. However, organization-enforced Auto-review and audit logs are Enterprise controls. A self-serve seat price does not include that governance package.

For Grok Bot, verify the actual account's plan and access before switching. Current pricing does not establish a complete country-by-country eligibility guarantee. Check regional and account access before paying for an alternative.

Are Dots and Grok Bot versions of the same idea?

Broadly, yes. Both offer ongoing assistance with context, a cloud computer, application access and work that can continue beyond a single conversation. The useful distinction is how you organize that work and what remains shared.

Grok Bot lets you organize named bots around different roles. However, its computer and apps documentation says those bots share files, browser sessions and command-line credentials within the user's environment. Client separation therefore needs more than separate Bot names.

Dots gives you one primary assistant inside ChatGPT for now. It organizes work into separate task conversations, including tasks on a connected computer. For your buying decision, the useful question is whether that arrangement fits the projects you want it to manage.

Both also reach beyond a chat window. A Dot has a separate cloud computer and browser, with optional local-computer access. Grok Bot combines its cloud environment with connected applications and browser work. Your particular client system still needs a supported, permitted connection and the right actions.

For a consultant using Slack, Notion, GitHub and client systems, make a list of the exact actions required: read a channel, retrieve a record, edit a document, send a message. Verify each action and approval path. Use that action list to assess the integration, rather than its place in a large app directory.

When Tasks in Gemini belongs on the shortlist

For an eligible individual whose work centers on Gmail, Calendar and Google files, Tasks in Gemini is a sensible starting point, especially when Pro is already paid for. Its native workflows can work with Google applications and produce documents and other files.

Google also supports non-Google connections. Its custom connected-app instructions describe MCP connections, but currently limit that route to the US and English. Writes through those custom connections require manual confirmation. That distinction matters for an international reader planning an unattended workflow across non-Google apps.

Schedules can run at set times, when a Gmail filter matches, or when a monitored topic changes. Google says run times are approximate and that usage limits or busy periods can prevent or delay execution. Google's pages also disagree on whether schedules run while your device is off. The schedules guide says Gemini runs in the cloud and keeps working; the main task guide says schedules will not run if your device is off. Test a schedule with your device off before relying on it overnight.

Local files create another boundary. On Mac, connected folders provide an explicit local access route. Cloud work and work dependent on your own computer are different: do not assume a local browser or folder task can finish while that computer is unavailable.

Who approves the overnight work

The approval point often matters more than the feature list. A research task may finish with a draft awaiting your approval. Include that handoff in the job you assign.

For Dots, OpenAI limits proactive background research to reading and private notes. Those research tools cannot send messages, change plugin content or operate a browser. Authorized background tasks follow the usual action rules. For actions, Custom Rules let you choose, per kind of action, whether the Dot acts without asking, acts only when you explicitly asked for it, asks first, or hands the step to you. Asking for a draft does not authorize sending it, and rules cannot override OpenAI's built-in safeguards.

Grok Bot's approval settings pair Auto Review with Ask first and Allow automatically rules, and Ask first wins when both match. An approval request from work that started without you, such as a routine, a trigger or a message from another Bot, expires after about 10 minutes and the action does not run. Approvals in a chat you are participating in wait for your response. Unattended work may therefore need a retry in the morning. xAI describes Auto Review as model-based, so treat it as a second check rather than a guarantee.

Google says Gemini is designed to ask before it sends communications, modifies your data, makes purchases or submits web forms, and it asks before each browsing task. Edits to shared Docs, Sheets and Slides need your review. Google names one exception: Gemini can make bulk changes to private items in Google Tasks without asking. It also warns that if a schedule runs while you are offline, you may not be able to stop an unintended action. Writes through custom connected apps always need manual confirmation.

A website's access rules still apply when you authorize an agent. A blocked browser workflow may require another supported route or a manual step. Choosely's guide to always-on agent permissions covers the wider decision.

Check what stays behind, and what trains

Disconnecting an app stops a connection. Information already retrieved can remain in memories, activity records or files, with separate deletion controls.

For Dots, the privacy FAQ says disconnecting leaves learned context in place, and individual Dot memories cannot be removed one at a time. Deleting the Dot clears its own context; files and other ChatGPT or Codex conversations it created are stored separately. On personal plans, the “Improve the model for everyone” setting governs eligible conversations and work. Business, Enterprise and Edu data is excluded from model training by default.

Tasks in Gemini requires Keep Activity on. Google's Privacy Hub says activity under that setting is used to improve services, including training generative AI models, with human review of a subset. Reviewed chats are disconnected from your account and can be retained for up to three years after activity deletion. Connected-app data also has specific processing terms. Turning Task mode off removes remote-browser and remote-computer data and pauses tasks and schedules; chats remain.

For Grok Bot, training choices follow Cursor's privacy settings, and Legacy Privacy Mode is unsupported. Cursor manages model selection, with no customer-facing model picker and a serving mix that can change. Bot computers run in the US, which alone does not establish a contractual US-only processing guarantee. Deleting a Bot leaves shared files and browser sessions, so project cleanup also involves removing files and signing out of sites.

Settle the account type, processing terms and deletion requirements your work needs before comparing monthly prices.

When your existing plan is enough

If your request is “give me a briefing every Monday,” start with ordinary scheduling. Google's scheduled actions are distinct from Tasks in Gemini and are rolling out to personal accounts and qualifying Workspace users.

OpenAI's scheduled tasks cover recurring work on Free and paid plans, and Plus can run event-triggered tasks in Work that respond to supported Gmail, Slack or GitHub activity. For those workflows, check the access you already have before upgrading to Pro.

An ongoing agent becomes more relevant when you need persistent project context, several steps across applications, or useful follow-up between explicit requests. Even then, define the output before buying: where it should appear, which sources it may use and which changes need approval.

Also check the meter. Grok Bot uses a shared weekly allowance with additional usage available. Gemini remains subject to usage limits. Dots launched with extended introductory usage, but OpenAI's launch announcement and release notes describe allowances differently. Budget against the current account allowance and recheck the terms when the introductory period ends.

The Choosely verdict

Start with the eligible capability you already pay for. For personal Google workflows, that may be Tasks in Gemini. For an eligible ChatGPT Pro user, it may be the included Dot. Buying both needs a specific unmet job.

Grok Bot is a serious alternative when a new subscription is necessary. Its lower current entry price makes it worth comparing with a Pro upgrade, particularly for readers who want role-based ongoing help. Shared credentials, approval interruptions and usage limits still need to fit the work.

Stay with ordinary scheduled tasks when the job is simple. The strongest reason to pay more is a clearly defined workflow your current tools cannot cover, with acceptable permissions and a manageable bill. A more persistent assistant is useful only if it produces work you can actually use.