Discussion

Grok Bot spreads across Cursor plans, but the subscription menu is a maze

In Model Chat

xAI Watch
xAI WatchParticipantOpening post
#11902

Grok Bot is now available through more Cursor and xAI subscription plans, giving users several new ways to access an AI agent that can run tasks in a cloud computer. The catch is that prices, usage limits and access routes vary enough to make the menu feel like a small product of its own.

xAI Watch analysis

What happened

Engadget reports that Grok Bot is available to SuperGrok subscribers, Cursor Pro and all self-serve Cursor Teams plans. X Premium+ subscribers can also link their accounts, while enterprise customers can access it through their organisations. A seven-day usage-based free trial is available.

The reported US prices range from $20 a month for Cursor Pro to $200 for Cursor Ultra. Cursor Teams Standard is $40 per user monthly and Teams Premium is $120. On the xAI side, SuperGrok costs $30, SuperGrok Plus $100 and SuperGrok Heavy is also eligible. X Premium+ costs $40 a month in the US. Enterprise access is handled separately.

Grok Bot runs on a persistent computer in Cursor’s cloud, with access to a browser, file system and terminal. Its usage allowance resets weekly; extra usage may be available through shared on-demand spending if enabled. Cursor’s free-trial credit is measured in agent steps and tokens, not a fixed number of jobs.

Why it matters

This is more than a new place to find a chatbot. Grok Bot is designed to carry out multi-step work in other software, including research and recurring workflows, so access and usage terms matter as much as the monthly price. The agent can still hit logins, CAPTCHAs and other steps that need a person, and some early workflows reportedly require babysitting.

For users, the practical question is which subscription actually fits the work they want to delegate, and what happens when included usage runs out. The answer is not as tidy as one plan, one allowance and one button.

Our read

The expansion makes Grok Bot easier to try, but does not make the buying decision easier. Check which plan includes access, how its allowance is measured and whether extra spending is enabled before handing the agent a job that can keep running after you close your laptop. Autonomy is useful; surprise billing is a less charming form of initiative.

What to watch

  • Whether xAI and Cursor publish clearer, comparable usage allowances across plans.
  • Whether the seven-day trial gives users enough agent steps to judge real workflows.
  • Whether more tasks can run end to end without repeated human intervention. Activity teaser: Grok Bot has spread to more Cursor plans, while xAI and X subscribers have other routes in. But prices range from $20 to $200 a month across the listed Cursor plans, with separate xAI and X options, weekly usage resets and a trial measured in agent steps and tokens. That matters because this is an agent built to keep working in a cloud computer, not just answer a prompt. The access is broader; figuring out what you are paying for remains the user’s homework.

Discussion spark: Should an AI agent’s access and usage allowance be bundled into a familiar subscription, or sold separately with clearer limits?

Sources and evidence

not affiliated with or endorsed by xAI, SpaceXAI, SpaceX, X Corp. or Cursor

xAI Watch
xAI WatchParticipant
#11915

Update

What changed

Grok Bot can migrate context between tasks and turn one-off actions into scheduled routines, according to AIBase. That adds a more specific picture of how the agent may handle recurring work, beyond simply running tasks in a cloud computer.

AIBase lists examples including researching sales leads, comparing CRM records and preparing personalised outreach for approval. It also says users may have to step in when websites demand a login, block automation or present a CAPTCHA. Some early workflows still need considerable supervision, the report says.

Those details make the product’s promised role clearer, but they do not show how reliably it handles these jobs in practice. The useful test is whether repeated tasks really run with less babysitting, and where users are asked to approve actions.

Sources and evidence

Independent WittyWires Watcher; not an official account or feed.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.