Cursor sells The X Subscription alongside legacy plans. The seat is priced, the pool refreshes weekly, and the free tier is still described rather than counted.
Plans table from Cursor's models and pricing page as it printed on 2026-11-03: The X Subscription lineup, Teams seats and annual billingCursor docs, models and pricing · Captured Nov 3, 2026
Individual developers and small teams who have already decided to adopt an AI-native editor, and who want a pricing page that states the overage mechanism instead of hiding it.
Not for
Anyone planning a budget from the free tier, anyone who needs included third-party model usage on a new subscription, or anyone who assumed a bring-your-own-key setup removes Anysphere from the path of their requests.
A coding agent whose pricing page changed shape in late 2026: new subscriptions get one weekly pool with prepaid top-ups, legacy plans keep two monthly pools with on-demand billing, and neither publishes how much usage you actually get.
Best for
Individual developers and small teams who have already decided to adopt an AI-native editor, and who want a pricing page that states the overage mechanism instead of hiding it.
Not for
Anyone planning a budget from the free tier, anyone who needs included third-party model usage on a new subscription, or anyone who assumed a bring-your-own-key setup removes Anysphere from the path of their requests.
We may earn a commission if you subscribe through this link. It never changes a score or a ranking. Read the disclosure.
The pricing and usage docs answer the follow-up questions in writing: what a plan includes, what happens at the limit, and what extra usage costs, all on dated pages
A genuine free plan that needs no credit card, even though the allowance is described rather than counted
The vendor publishes the overage mechanics plainly for both lineups: prepaid top-ups with an optional Auto Top-up on the current plans, and on-demand at API rates with no markup on legacy plans
Privacy Mode is a written guarantee with a zero-retention commitment that names its exceptions, not a marketing phrase
What to weigh
No plan publishes how much included usage you get: the current pool is described only as a multiple of Plus (4x, 15x, 50x), and Plus itself is never given a figure
New subscriptions no longer include third-party model usage at all: those models are billed separately at API list price from prepaid credit, and only on Premium and above
The pool now refreshes weekly rather than monthly, so a heavy week is stopped sooner, and top-up credit is charged when you buy it rather than billed in arrears
Bring-your-own-key requests still route through Anysphere's backend, and switching editors is part of the purchase, so the exit is heavier than an extension's
01What the vendor publishes
Read from the vendor's own pages on the dates listed at the bottom of this page. Where a figure is not published anywhere we could find, the row is absent rather than guessed.
One shared pool for Cursor models, refreshes weekly, shared with Grok7
Third-party models, current lineup
Not included; billed at API list price from prepaid credit, Premium and above only8
Extra usage mechanics, current lineup
Prepaid top-ups, charged when bought; Auto Top-up off until enabled; credit expires 365 days after purchase9
Legacy plan usage mechanics
Two monthly pools (Cursor Models, Other Models); opt-in on-demand billed in arrears at API rates with no markup10
Bring your own API key, individual plans
Provider bills you directly; requests still route through Anysphere's backend11
Privacy Mode
Customer Data not used for training by Cursor; zero data retention agreements with model providers12
Privacy Mode off
Codebase data, prompts and editor actions may be used to train models13
Teams entry price, current lineup
$30, $100, or $200 per seat/mo, matching the seat's tier; Plus is not available for teams14
Cursor Token Rate
$0.25 per million tokens on Teams and Enterprise third-party requests15
Cursor prices the entry in plain dollars and now runs two pricing systems at once. What it still does not price is the amount of usage a seat includes, and that gap is the part of the cost a buyer cannot predict from the page.
Every price here was read off cursor.com on 2026-11-03, and reading that page today is a different exercise than it was in September. The plans table in the docs now describes The X Subscription: Free at $0, Plus at $8, Premium at $30, Super at $100 and Ultra at $200 per month, sold through X Subscription Corp, with Grok models featured across the tiers and X benefits bundled into the paid cards. The lineup this review described in September, Hobby, Pro, Pro+ and Ultra, still exists but is documented as legacy: it is for existing subscribers, it keeps two monthly pools and on-demand usage, and Anysphere's legacy-plans doc says it "keeps working as it is", with switching optional and permanent.
Both lineups are live at once, which is why this review reads both. If you subscribe today you get the new mechanics; if you already pay, nothing changes until you choose to switch.
On the current lineup the pricing page describes included usage only as a multiple of Plus: Premium is 4x, Super 15x, Ultra 50x. It never sizes Plus itself. The pool is one shared allowance for Cursor's own models, shared with Grok, that refreshes weekly, so the unit of planning is a week rather than a month.
Third-party models are the bigger change. They are not included on the current lineup at all: the pricing page's footnotes and FAQ state that they are available in Cursor billed separately by usage, at API list price, from prepaid credit, on Premium and above. On legacy Pro, Pro+ and Ultra, the Other Models pool still includes third-party usage drawn at each model's API price. A buyer who routes most work to third-party models is therefore choosing between two materially different cost structures depending on which side of the lineup split they land.
When the weekly pool runs out on the current lineup, extra usage comes from top-ups: prepaid credit charged when you buy it or when an automatic top-up runs, not in arrears. Auto Top-up is off until you turn it on, credit currently expires 365 days after purchase, and a monthly additional usage limit can cap what paid credit can consume. On legacy plans, on-demand usage is opt-in for individuals, billed monthly in arrears at the same API rates with no markup, with a spend limit whose enforcement the docs note is not instant.
Evidence · Cursor help docs, usage-based charges · Usage-based charges help page, captured 2026-11-03: prepaid top-ups on The X Subscription, opt-in on-demand on legacy plans · Captured Nov 3, 2026
The figure that is still absent is the one a buyer needs first: nowhere on the pricing page or the models and pricing docs is the included allowance stated as a dollar or token figure. The docs' closest guidance is by tier, "Premium suits new Agent users, Super suits daily Agent users, and Ultra suits power users running several agents or Automations", which names an audience rather than an amount.
The Free plan is genuinely free and needs no credit card. Its card says Limited AI usage, applied to agent and chat requests, which is the whole sentence. The word limited is not followed by a number on the pricing page, and the docs describe free access in terms of features rather than an allowance.
That is a factual statement about the page, not a criticism of the plan. It means a reader cannot answer from the vendor's own material whether the free tier is enough for a week of real work. If the answer is important to you, the only way to get it is to run the plan and watch the usage dashboard.
Anysphere publishes the data arrangement in one dated document with two states. With Privacy Mode on, customer data is not used for training by Cursor, and Anysphere says it maintains zero data retention agreements with all providers, with model providers not storing or training on your data. The same sentence carries qualifications: model providers may still run risk classifiers, and data that triggers an abuse detector may be stored for investigation under the provider's retention policy, and non-ZDR models are designated as such or require an admin opt-in. With it off, the same document states that codebase data, prompts, editor actions and snippets may be used to improve AI features and train models.
None of that is hidden, and none of it is permanent by default. Privacy Mode is a setting, available to any user, and the guarantee is attached to the mode rather than to the product, with the exceptions above stated in the same paragraph as the promise.
Two facts shape how hard it is to leave. First, requests made with your own API key still pass through Anysphere's backend for final prompt building, even though on individual plans your provider bills you directly and those requests do not draw from the plan's usage pools. Second, adopting Cursor is adopting an editor, not a plugin on the one you already use, so a decision to leave is a decision to migrate your editing workflow as well.
The lineup split adds a wrinkle the old review did not have: switching to The X Subscription is permanent, and it moves billing from cursor.com to grok.com, cancels eligible subscriptions at checkout with a prorated refund, and converts two monthly pools into one weekly one with third-party models no longer included. Anysphere's own doc says comparable plans include at least twice as much Cursor-model usage, which is the vendor's framing of the trade; the third-party allowance it drops is ours to state.
Taken together they mean the portability of your prompts is not the same as the portability of your subscriptions. The AI-native editor choice is closer to a platform change than to an uninstall.
The five components and their weights are published on our methodology page. Here is what this review put under each of them rather than a number with nothing behind it.
Fit for the stated job, 30 percent. Cursor is an editor that is also a coding agent, which is the whole of what this category is about rather than a corner of it.
Cost to get started, 25 percent. A real free plan with no credit card, an entry price of $8 on the current lineup that is printed plainly, and overage mechanics stated in writing for both lineups. Held back by the missing included-usage figure, which moves the cost question to after you subscribe.
What the vendor commits to, 20 percent. The Privacy Mode guarantee, the zero data retention commitment, and spending controls with published behavior on both lineups keep this component strong.
Portability, 15 percent. Own-API-key requests still route through the vendor, the product demands an editor change, leaving is a migration rather than an uninstall, and the switch to the new lineup is one-way.
Documentation quality, 10 percent. The pricing, usage, overage and legacy-plan pages answer the follow-up questions a buyer asks, dated and with the edge cases named, including a full page devoted to the lineup split.
There is no score here for output quality, because we did not measure it and we have not found an independent benchmark of coding-agent output we would be willing to rank on. There is also no figure for the amount of included usage, on either lineup, because the current docs describe pools as multiples of a Plus allowance they never size, and the legacy docs describe two pools without printing an amount for either. Publishing a number to fill the row would be worth less than the empty space.
Only to the extent that the vendor describes it. The Free card says Limited AI usage with no number beside it, and the docs describe the paid pools as multiples of Plus without ever sizing Plus itself. What you can rely on is that a paid seat replaces the question of whether you may run extra usage, not that you can predict the total before spending.
It matters for both price and mechanics. New subscriptions are on The X Subscription: Free, Plus $8, Premium $30, Super $100, Ultra $200, one weekly pool, prepaid top-ups, no included third-party models. Accounts that subscribed earlier may stay on legacy plans: Hobby, Pro $20, Pro+ $60, Ultra $200, two monthly pools, opt-in on-demand, third-party models included on Pro and above. Anysphere's docs say legacy plans keep working as they are and switching is optional and permanent, so the legacy table on its own page, not the current pricing page, is the source for those prices.
No, not entirely. Anysphere says individual requests using your own key are billed directly by your provider and do not draw from the plan's usage pools, but the help docs also state that even with your own key the requests still go through Anysphere's backend for final prompt building. The two statements are both true and describe different parts of the path.
With Privacy Mode on, Anysphere writes that customer data will not be used for training by Cursor, and that it maintains zero data retention agreements with all providers, with model providers not storing or training on your data. Those promises carry exceptions in the same document: providers may still run risk classifiers, flagged data may be held for investigation under a retention policy, and non-ZDR models are designated or require an admin opt-in. With it off, the same page says your codebase data, prompts and editor actions may be used to improve AI features and train models. It is a mode you choose, located in settings, and the differences are written down.
Depends on the lineup. On The X Subscription, extra usage comes from prepaid top-ups, so you are charged when you buy credit rather than billed after the fact, and Auto Top-up is off until you turn it on. On legacy plans, on-demand usage must be enabled explicitly and a spend limit can be set, with the docs noting that enforcement is not instant. In both cases the default is that nothing extra happens until you choose it, which is also the catch: the choice to allow more usage is yours and it is visible ahead of the bill.
04Source ledger: every claim on this page with its document and the date it was checked
Where these numbers came from
Every figure above is the vendor's own published number. Each row names the document it was read from and the day a human opened it.
Source ledger: every claim on this page with its document and the date it was checked