Cursor pricing in October 2026: every plan, what it includes and whether it is free

How much does Cursor cost in October 2026?

Cursor changed how it sells plans in 2026, so most pricing pages you will find through search are out of date. The tables below come from Cursor's own Models & Pricing docs and pricing and plans help page, read in October 2026. Cursor now says it is sold as part of The X Subscription, one subscription for Grok, Cursor and X. New customers start there. People who subscribed earlier are on what Cursor calls legacy plans.

The multipliers are relative to Plus. Premium has four times Plus's included usage, Super fifteen times, Ultra fifty times. Cursor's tables describe usage as multiples of Plus rather than as a fixed token count, so the multipliers are the way to compare plans.

Is Cursor free?

Yes, with limits. Cursor's help page says the Free plan gives you access to core features with limited usage: Agent, Chat and Tab completions with Cursor models (Grok and Composer). The older Hobby plan works the same way. Cursor's docs describe Free usage only as limited, so you will find the ceiling by using it.

What Free leaves out matters more than the cap. Cloud Agents, Bugbot and Origin are not included, and neither are third-party models. If you want Claude or GPT inside Cursor, you need Premium or above. Projects, which run on Cloud Agents, are not available on Free either.

• Good enough for: reading an unfamiliar codebase, a few small edits, and finding out whether the editor suits you.
• Not enough for: building a whole app in one sitting with an agent. A long agent session can use a lot of the allowance.
• Students: the old student discount ended for new sign-ups on June 25, 2026. The details are in the guide to Cursor for students.

A practical way to test Free: pick one small idea, describe it in a single prompt and see how far the allowance takes it. The board has ideas small enough for that, each with a build prompt you can paste in. If the allowance runs out before the first version works, you have your answer about which paid tier you would need.

How do Cursor's usage limits work?

Each paid plan includes one pool of usage for Cursor's own models. According to Cursor's docs, the pool is shared across Agent in the editor, the CLI, Cloud Agents, Grok Bot and the Grok apps, and it refreshes weekly, so one heavy day does not use up your month. Your remaining usage and next reset date show in the editor settings and on the usage page at grok.com/dashboard.

Different models draw from the pool at different rates. These are the per-million-token rates Cursor lists for its own models. They apply when you buy usage beyond the weekly pool, and they show how fast each model eats allowance.

A worked example makes the rates concrete. Suppose one agent step reads 100,000 input tokens and writes 5,000 output tokens on Grok 4.7. At the listed rates that is 0.1 x USD 2 + 0.005 x USD 6 = USD 0.23. The same step on Composer 2.5 is 0.1 x USD 0.5 + 0.005 x USD 2.5 = USD 0.0625. Long agent runs make many such steps, and the docs note that Grok 4.7 context above 256k input tokens is billed at twice the standard rate. Fast mode costs more: on Grok 4.7 it doubles the rate.

The point of the arithmetic is the ratio, not the cents: on these rates Composer 2.5 costs a quarter of Grok 4.7 for the same step. If you hit the weekly limit often, the cheaper model is the first lever to try.

What do Claude, GPT and Gemini cost inside Cursor?

They are not part of the included pool. Cursor's docs say that on Premium and above, third-party models are available at the model's API list price from the first request, paid from prepaid credit. Free and Plus do not include them, and Plus does not let you bring your own API key.

Prepaid credit works like this, per the docs:

• Top-ups add credit to your subscription whenever you choose.
• Auto Top-up refills credit when your balance drops below a threshold you set.
• Monthly additional usage limit caps how much paid extra usage you can consume each month, however much credit you hold.
• Credit pays for third-party models and for Cursor models beyond the weekly pool, at the rates above.

When the weekly pool runs out you can wait for the reset, keep going with prepaid credit, or move to a higher tier. Cursor says requests are never downgraded in quality or speed. Set the monthly limit before you turn on Auto Top-up, so a runaway agent cannot spend more than you decided.

Which Cursor plan should you pick?

Cursor's own guidance is that Premium suits new Agent users, Super suits daily Agent users, and Ultra suits power users running several agents or Automations. That matches how the tiers are built. Use the table as a first filter, then check the limits against your own habits.

If you are not sure, start one tier lower than you expect to need. Upgrades apply right away, while downgrades apply at the end of the billing period, so moving up costs you nothing in waiting and moving down is slow.

What does Cursor cost for a team?

Teams are priced per seat. Each member gets a Premium, Super or Ultra seat at the same price and with the same weekly usage as the matching individual plan, and you can mix seat tiers on one team. Plus is not available for teams. Team plans add centralized billing, a team marketplace for internal rules, skills and plugins, usage analytics, privacy-mode enforcement, custom Chat data retention and SAML/OIDC SSO. Enterprise adds pooled usage across Grok and Cursor, invoicing, SCIM and granular spend controls.

One extra charge applies only to teams. On Teams and Enterprise, eligible third-party model requests carry a Cursor Token Rate of USD 0.25 per million tokens on top of the model's API price. Individual plans do not pay it, and neither do Cursor's own models. A list of cheap models, including several Mini, Nano and Flash variants, is excluded from the rate.

A worked example for five people: five Premium seats cost 5 x USD 30 = USD 150 a month. The same team on the legacy Teams Standard seat at USD 40 would pay USD 200. Cursor says each paid X Subscription tier includes at least twice the coding usage with Cursor models as the legacy plan it replaces, but it drops the included third-party allowance, so a team that relied on Claude inside the old plan should compare both bills.

What happened to Cursor Pro and the older plans?

They still exist for existing customers. Cursor's docs say legacy plans keep working as they are, and switching to The X Subscription is optional and permanent. Legacy plans use two monthly usage pools (Cursor Models and Other Models) and on-demand usage billed monthly in arrears. The new plans use one weekly pool and prepaid top-ups.

Cursor calls the match the closest tier, not an identical one. The biggest practical difference is third-party models: legacy Pro, Pro+, Ultra and Teams include an Other Models pool that covers Claude, GPT and Gemini, while the new plans bill them from prepaid credit.

1. If you are on a legacy plan and happy, do nothing. Your price and features continue unless you change or cancel.
2. If you rely on third-party models, compare your last three months of Other Models usage with API list prices before switching.
3. If you pay through the Apple App Store, remember Cursor cannot cancel or refund that subscription for you. Cancel it in your device settings or Apple keeps billing you.
4. If you do switch, Cursor cancels the old subscription at checkout and refunds unused time on a prorated basis. The switch cannot be undone.

Where you manage billing depends on the plan. The X Subscription is managed at grok.com/dashboard/billing, and legacy Cursor plans at cursor.com/dashboard/billing.

Is annual billing cheaper?

Yes. Cursor lists annual prices for each individual paid plan, and in every case the yearly price equals ten months of the monthly price.

That is about 17 percent off. The catch is commitment: tools in this category change fast, and a plan that fits today may not fit in six months. Pay monthly until your usage pattern is stable, then switch.

How do you keep a Cursor bill under control?

• Match the model to the task. Use the cheaper model for renames, boilerplate and small edits, and the stronger one for planning and hard bugs.
• Plan before you build. Cursor's Plan Mode produces a reviewable plan before any code is written. A wrong plan costs a few messages. A wrong build costs a whole run. The guide to the Cursor agent covers it.
• Keep chats short. Every chat shares a fixed context window, and long chats carry more tokens on each step. Start a fresh chat for a new task.
• Write rules once. A short rules file saves repeating instructions in every prompt. See Cursor rules.
• Set the monthly additional usage limit. It is the one number that caps your worst case.
• Watch the usage page in your settings after your first heavy day, not at the end of the month.

Before I pay for Cursor:
[ ] I used Free for at least three real tasks
[ ] I know whether I need Claude, GPT or Gemini (Premium or above) or whether Cursor's own models are enough
[ ] I know whether I need Cloud Agents (Premium or above)
[ ] I chose monthly billing, not annual, until my usage is stable
[ ] I set the monthly additional usage limit before turning on Auto Top-up
[ ] I wrote down the date I will review this plan: ______