Finding out whether it can do your work at all
Premium
You have read what it claims and you want to watch it try one real task before committing a card to anything.
$30/mo
$25/mo billed annually
There is a trial, and it is not the kind you think. It is a usage credit rather than a stretch of days — the seven-day window is the outer bound, not the meter — and it is drawn down by agent steps and tokens rather than by messages. The vendor states outright that one large job can consume all of it in a single run and that spent credit is never restored, so the trial rewards a small, tightly scoped first task. When it runs out, Premium is the cheapest tier that includes Grok Bot; the two tiers below it, Plus and Free, do not. One warning before you press anything: ending the trial early deletes whatever credit is left, and it cannot be claimed again.
You already code in Cursor every day
Super
The editor is open eight hours a day, and you want bots running alongside it on the same subscription.
$100/mo
$84/mo billed annually
This is where the new plan changes the arithmetic. On the old Cursor plans, Grok Bot ran on a weekly grant of its own and the vendor said Bot work did not touch your Cursor usage. On The X Subscription there is one weekly pool, and Grok Bot, Cursor's agent, the CLI, cloud agents and the Grok apps all draw on it, at rates that differ by product. A heavy coding week therefore leaves less for the bots, and a busy week of bots leaves less for the editor. Super is the tier the vendor recommends for daily agent users, and the honest way to choose between it and Premium is to watch the one meter for a week.
One person, several bots, running all week
Ultra
You have got past the demo, you have three or four standing routines, and you keep hitting the weekly ceiling.
$200/mo
$167/mo billed annually
This is the top of the individual ladder, and the pricing is answerable only in ratios. The Grok Bot page says Super carries 3.5x and Ultra 12x the Premium pool; the vendor's team pricing docs put the same seats at 3.75x and 12.5x. No page gives the Premium pool itself as a quantity, so the question a reader at this level actually has — what a week of four routines costs — cannot be answered from anything published. Past the pool, extra work is paid from prepaid credit, billed on token cost. The meter in the app is the only place either number becomes real.
Putting it on a whole team next week
Teams
Several people, one invoice, and no appetite for chasing individual subscriptions through expenses.
$30/seat/mo
Straightforward by the standards of this page. Teams seats are the same three tiers at the individual prices, they can be mixed on one team, and every paid seat includes Grok Bot with nothing for an admin to switch on. Two facts matter to whoever owns the budget. Each seat has its own weekly pool, which teammates cannot share — only Enterprise pools usage across people. And what the team does share is one prepaid balance for everything past those pools, with one monthly cap set by admins; when either runs out, paid usage stops for the whole team.
Somebody has to sign off on it first
Enterprise
A security review stands between you and a rollout, and it will ask where the data goes and what the thing is allowed to touch.
Contact sales
This is the only route where the answers are configurable rather than fixed, and the controls that exist nowhere else are precisely the ones a review asks for: an organisation-wide switch, network controls, team setup scripts, access and audit controls for Grok Bot. Go in with two facts already in hand. Every bot on an account shares one persistent computer, so isolation is per user, not per bot, and separate bots are not a security boundary. And the vendor manages model selection, so the model allowlist a review will fixate on is a policy you set, not a picker you control.