Cursor Pricing in 2026: The X Subscription Plans vs Legacy Pro, Pro+ and Ultra

The short answer

The rest of this guide is built to help you choose rather than to repeat Cursor’s tables. All prices and plan terms come from five first-party Cursor pages: the Pricing and plans help page, the Models and Pricing docs, the Legacy Cursor plans docs, the Usage and limits help page and the cursor.com/pricing page, all read on October 30, 2026. The cost arithmetic, the price-per-unit table and the break-even reasoning are ours, built from those published numbers. We did not subscribe, run the switch flow or test any plan, so everything about how the plans behave in practice rests on Cursor’s documentation.

Prices change. Cursor’s pricing page notes that prices are subject to change and exclude taxes, so confirm on the page before you pay. For background on the model that sits at the center of the new plans, see our Grok 4.5 and Cursor launch guide.

What changed: Cursor inside The X Subscription

Cursor’s pricing help page opens with one sentence that explains the whole reshuffle: Cursor is part of The X Subscription, one subscription for Grok, Cursor and X. New customers choose a plan on The X Subscription. If you subscribed to Cursor before it launched, you are on a legacy Cursor plan, and Cursor says it keeps working as it is.

That is the extent of the relationship we state. Cursor’s models page adds two more first-party details: it labels Grok 4.7, Grok 4.6 and Grok 4.5 as jointly trained by Cursor and SpaceXAI, and it says Cursor is sold as part of The X Subscription alongside Grok and X. The pricing page adds that subscriptions are sold directly through X Subscription Corp, and that Cursor does not authorize resellers. Beyond those statements, ownership, revenue sharing or corporate structure is not something Cursor’s pricing pages explain, and we do not speculate about it.

What matters for your wallet is the practical consequence. There are now two billing systems that share the Cursor name. The new one bundles Grok and X benefits, meters Cursor’s own models in a weekly pool shared with Grok, and bills third-party models separately. The old one meters usage in two monthly pools and includes third-party models in one of them. Billing for the new system lives at grok.com/dashboard/billing, and billing for legacy plans stays at cursor.com/dashboard/billing.

The current plans

These are the plans a new customer sees, with prices exactly as Cursor’s help page and docs list them on October 30, 2026. Included usage is expressed relative to Plus, and Cursor does not publish an absolute token or dollar figure for the pool in the pages we read.

Cursor adds that every paid plan includes a weekly pool of usage for Cursor models, which are Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5, and that the pool is shared across Cursor and Grok. Premium adds Cloud Agents and Automations, code hosting on Origin, code review with Bugbot, third-party models at API list price, and Grok Bot. Super adds priority model access at peak times. Ultra adds earliest access to new products and models. The Free plan gives you Agent, Chat and Tab with limited usage on Cursor models, along with standard X.

Annual billing is a modest discount. Plus at $80 a year is $16 below twelve months at $8, and Premium at $300 is $60 below twelve months at $30. Across the four paid tiers the annual price works out to ten months of the monthly price, by our arithmetic. Cursor’s docs also note you can switch monthly and annual billing when you change plans at grok.com/dashboard/billing.

The legacy plans

Legacy plans are what Cursor sold before The X Subscription. They are not available to new customers, but existing subscribers keep them until they switch or cancel. Cursor’s docs map each one to a comparable new tier, and say plainly that comparable means the closest tier, not an identical one.

The last column is our arithmetic on Cursor’s listed prices, not a Cursor claim. It shows an asymmetry worth noticing: on the mapped tiers, individual Pro and Pro+ users would pay more list price on the new plans, while legacy Teams seats would pay less. Cursor’s docs justify the individual increase this way: each paid plan on The X Subscription includes at least twice as much coding usage with Cursor models as the legacy plan it replaces, and it adds Grok and X benefits. In the same breath the docs warn that the new plans drop the included third-party model allowance, and tell you to compare both before you switch.

Legacy Pro, Pro+ and Ultra include both usage pools, unlimited Tab completions, Bugbot, Cloud Agents, Automations, Origin and the Cursor SDK, with Fast mode and Grok effort levels available on all three. Pro+ and Ultra include more of each pool than Pro, but the legacy docs we read do not publish the pool sizes. Start, the India-only plan, includes the Cursor Models pool but not the Other Models pool, on-demand usage, Bugbot, Auto, Automations or the Cursor SDK.

How the two systems differ

The headline prices hide the mechanics that decide your actual bill. Cursor’s legacy plans page lays out the differences in one table, which we reproduce in condensed form.

Weekly versus monthly, in practice

Cursor says a weekly refresh means one heavy day does not use up your month. That helps if your work is spiky, because a sprint on Monday cannot leave you stranded for three weeks. It cuts the other way if your usage is even and you benefit from monthly pooling, or if you save usage for a crunch. Unused included usage does not roll over on either kind of plan, per the usage and limits page, so the weekly reset does not penalize you more than the monthly one on that count.

On-demand versus prepaid

On legacy plans, going past a pool is optional pay-as-you-go. For individuals it is off until you turn it on under Spending, and it is billed monthly in arrears at the same per-model rates as included usage. On The X Subscription, the equivalent is prepaid credit you buy first, with a monthly cap you set. Cursor says credit stays available until used or until it expires, currently 365 days after purchase. For budgeting, prepaid credit is the more predictable of the two, because you cannot be invoiced for more than you loaded and capped.

What Plus really is

Plus is the headline-grabbing $8 plan and also the easiest to misread. The pricing help page and models page describe it consistently: Agent with Cursor’s own models (Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5) at fixed settings. Fast is off, effort is medium, and Grok 4.7 uses the 256k context window. It also includes Grok Chat, Voice, Imagine, Build and Grok API access. It does not include Auto, Cloud Agents, Automations, Bugbot, Origin, third-party models or your own API keys.

Two of those exclusions deserve emphasis. First, third-party models are not merely un-included on Plus, they are unavailable. Cursor’s models FAQ says plainly that you cannot use third-party models on Plus, and that you need Premium or higher for them and for the ability to choose Fast mode and effort levels. So Plus cannot be a cheap front door to Claude if you top up credit; it is a Cursor-models-only plan. Second, no bring-your-own-key. If you hoped to use Plus as an $8 editor shell around your own Anthropic or OpenAI key, Cursor’s docs list your own API keys as not included.

There is also a wording difference between two Cursor pages. The pricing page footnote for Plus says only Grok models are available on the plan, while the docs list Composer 2.5 among the Cursor models that Plus includes. Composer is Cursor’s own model rather than a Grok model, so the pricing page footnote is looser than the docs. We follow the docs and flag the mismatch.

Finally, Cursor’s own recommendation tells you who Plus is for. The pricing page says Premium is recommended for new AI users, Super for daily agent users, Ultra for power users, and Plus for social creators with limited AI usage. The Plus feature list on that page is mostly X benefits: a Premium checkmark, fewer timeline ads, enhanced creator tools and advanced analytics. If you are a developer who simply wants a cheap Cursor, Plus fits only if fixed-setting Cursor models cover your work.

Price per unit of included usage

Cursor does not publish the absolute size of the weekly pool, but it does publish multipliers against Plus. Dividing each plan’s monthly price by its multiplier gives a price per unit of included Cursor-model usage. This is our arithmetic, assumes the multipliers are exact, and ignores everything else in the plans.

Two points follow. Plus is not much cheaper per unit than Premium, so the $8 saves you money only if you would otherwise leave much of a 4x pool unused. And the jump to Ultra halves the unit price compared with Plus. If you are certain to burn the usage, higher tiers are better value per unit, which is consistent with Cursor’s recommendation that daily agent users take Super and power users take Ultra. If you are not certain, the cheaper tier with a spending cap is the safer experiment, since downgrades apply at the end of your billing period while upgrades apply right away.

Third-party models: what they cost now

This is where the new and old plans diverge most. On The X Subscription, third-party models are not part of any pool. On Premium and above they are available at the model’s API list price from the first request, paid from prepaid credit. The table shows a selection of rates from Cursor’s models page as of October 30, 2026, in dollars per million tokens. Names are Cursor’s own. We did not reconcile them with other vendors’ naming, and the table is not the full list, which runs to dozens of entries, many hidden by default in the picker.

Cursor’s own models are priced in the same table, and those rates apply when you buy usage beyond the weekly pool. Grok 4.7 lists at $2 input, $0.5 cache read and $6 output, with a Fast variant at $4, $1 and $12. Composer 2.5 lists at $0.5, $0.2 and $2.5, with a Fast variant at $3, $0.5 and $15. Cursor adds that Grok 4.7 above 256k input tokens bills at 2x standard rates, up to 500k.

What a task costs at list price

To make list prices tangible, here is our arithmetic for one agent task that reads 80,000 uncached input tokens and writes 20,000 output tokens. It ignores caching, which would lower input cost, and it ignores the extra tokens agents spend on reasoning and tool loops, which would raise it. Treat it as a floor, not a forecast.

Read the table against the plan gaps. The $10 list-price gap between legacy Pro and Premium equals roughly 14 Opus 5.5 tasks of this shape, or about 28 Sonnet 5.5 tasks, or about 5 or 6 Fable 5.1 tasks. The $22 gap between Plus and Premium equals about 30 Opus 5.5 tasks. These are not a recommendation to count tasks, only a feel for scale: on Premium, every task on a premium third-party model costs real money on top of the subscription. For ways to judge which of these models is worth that spend, see our best AI model for coding guide, which compares models on price and intelligence with dated scores.

Who should pick what

With the mechanics in place, here is the decision guide. It combines Cursor’s own recommendations with the arithmetic above. Where we infer rather than quote, we say so.

Should you stay on legacy Pro?

This is the question that drives most searches about the plan change, and Cursor’s documentation supports a precise answer even though it does not make the choice for you.

Start with what is free to do: nothing. Cursor’s help page says switching is optional, and that your legacy plan, price and features continue unless you change or cancel. You can also still move between legacy plans, from Pro to Pro+ for example, without switching. Cursor says upgrades take effect immediately, with any credit for your previous plan based on its unused included usage rather than days remaining, and downgrades apply at the end of the billing period.

Now consider the third-party usage on your account. The legacy docs say the Other Models pool is drawn down at each model’s API price, and that on-demand usage is billed at the same per-model rates as included usage. On The X Subscription, third-party models are billed at API list price from prepaid credit. Taking those statements at face value, a given Claude or GPT workload costs the same dollar amount per token on both systems. The difference is that on legacy Pro the first slice is covered by the pool inside your $20, and on Premium none of it is covered by the $30.

Three profiles come out of that reasoning. A heavy Claude or GPT user, who spends most of the Other Models pool every month, has the most to lose from switching and the least reason to do it. A Cursor-models user, who mostly runs Composer or Grok and rarely touches third-party models, has the most to gain, because the new plan includes at least twice the Cursor-model usage and adds the Grok and X benefits for a $10 increase. A mixed user should compare last month’s real numbers. If the third-party share is under $10 or so, the extra $10 may be the main cost of moving, and Cursor-model headroom and Grok may justify it. If it is well above, staying looks better.

One more reason to wait: you lose nothing by waiting, except the new tiers’ benefits, and you lose something permanent by switching. If your needs change later, Cursor’s docs describe the change as permanent and the pages we read describe no way back, so a wrong move cannot be undone by an easy toggle.

Switching is permanent: what happens

Cursor states it in two places. The legacy plans page says: switching to The X Subscription is optional and permanent. The pricing help page says: if you do switch, your Cursor subscription is cancelled with a prorated refund, you pick a plan on The X Subscription, and the change is permanent. Here is what the legacy docs list as changing at the moment you switch.

• Eligible Cursor and Grok subscriptions are cancelled at checkout, with a prorated refund for unused time. Grok Bot subscriptions bought in the iOS or Android app are cancelled and refunded for unused time too. A Cursor subscription billed through the Apple App Store must be cancelled by you in iPhone or iPad Settings.
• Any on-demand usage not yet invoiced is billed in one final invoice, and the switch waits until it is paid.
• Your usage moves from two monthly pools to one weekly pool for Cursor models, shared with Grok.
• Third-party models stop being included. They remain available in Cursor on Premium and above at API list price from prepaid credit.
• On-demand usage and its spend limit are replaced by top-ups and a monthly additional usage limit.
• Billing moves from cursor.com to grok.com/dashboard/billing.

A few edge cases from the same docs. When a paid Cursor account switches, the unify flow offers Premium and above, and Pro, Pro+, Teams Standard and Teams Premium default to the mapped tiers shown earlier. The Start plan can only be purchased while your account is on the free plan, so you cannot switch back to it directly from another paid plan. And if your Grok account holds more Grok Bot data than can be transferred yet, the flow asks you to try again later. Cursor’s docs point to cursor.com/unify for individual plans and cursor.com/unify/team for teams. We could not load the unify page when we tried, so we describe the flow only from Cursor’s documentation.

How to decide in ten minutes

1. Find out which system you are on. If you pay through cursor.com/dashboard/billing and see Pro, Pro+ or Ultra, you are on a legacy plan. If you manage billing at grok.com/dashboard/billing, you are on The X Subscription.
2. On legacy, open the Spending tab. Cursor says it shows real-time usage for both pools, remaining allowance and any on-demand charges. Write down last month’s Cursor Models usage, Other Models usage and on-demand charges.
3. Split your work by model. Count how much of last month went to Composer or Grok versus Claude, GPT or Gemini. Only the Cursor-model share is covered by the new plans’ weekly pool.
4. Price the third-party share at list. Use the per-million-token rates above, or the model table in Cursor’s docs, to estimate what the same work would cost from prepaid credit.
5. Add it up for each candidate. Premium cost is $30 plus third-party spend at list. Legacy Pro cost is $20 plus any on-demand overage. Plus cost is $8 and cannot run third-party models at all.
6. Check feature needs. If you need Cloud Agents, Automations, Origin, Bugbot or Auto-like routing, Plus is out. Note that individual plans have Cost mode only for Auto, and Cursor Router is for Teams and Enterprise.
7. Set controls before you start. On The X Subscription, set a monthly additional usage limit and decide whether to turn on Auto Top-up. On legacy, on-demand usage is off for individuals until you enable it, and you can set a spend limit under Spending.
8. If you are still unsure, do not switch. Staying costs nothing and keeps legacy terms. Switching cannot be undone by a toggle.

Teams and Enterprise

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

One cost line applies only to teams. On Teams and Enterprise plans, eligible third-party model requests include a Cursor Token Rate of $0.25 per million tokens on top of the model’s API price, covering additional usage and bring-your-own-key usage. Individual plans do not pay it. Cursor models, Cost mode requests and a published exclusion list do not carry it either, and the list includes Claude Haiku 5.5, GLM 5.3 Flash, GPT-5.6 Luna and several other small models. Cursor Router, which routes Auto requests among models in Balance and Intelligence modes, is launching for Teams and Enterprise, and Cursor says individual plans have Cost mode only.

If you are weighing Cursor against other agent IDEs on cost, our Cursor vs Antigravity comparison covers price and features side by side. If you are also looking at GitHub’s usage-based credit model, our GitHub Copilot AI credits guide shows how another vendor meters agent usage.

Caveats and unknowns

• Cursor does not publish the absolute size of the weekly pool, nor the dollar size of the legacy Pro Other Models pool, in any page we read. All comparisons use Cursor’s multipliers or our own arithmetic.
• We did not subscribe, run the unify switch flow or test any plan, and we could not load cursor.com/unify. How the switch behaves in practice is described only from Cursor’s documentation.
• The documents do not say whether Plus can buy prepaid credit for usage beyond its weekly pool. The top-up section says credit covers Cursor models beyond the weekly pool, without naming tiers.
• No page we read states an end date for legacy plans. That does not mean none exists; it means Cursor has not said so in these documents.
• Cursor’s docs spell the partner as SpaceXAI in the model table and as SpaceXSI once in the introduction. We use SpaceXAI, as the model table and our earlier coverage do, and flag the inconsistency.
• The pricing page lists Teams at $30 or more per user per month, while the docs give $30, $100 or $200 per seat. We use the docs figures.
• Model rates, promotions and model names change often. The GPT-5.6 Sol rate is described as promotional through November 21, 2026, so expect it to change.
• Our per-task cost table uses one made-up task shape with no caching and no reasoning overhead. Real agent sessions will differ, often by a lot.

Verdict

The new Cursor lineup is simpler on paper and harsher for people who live on Claude or GPT. The X Subscription gives you a clean ladder, Plus to Ultra, with a weekly pool for Cursor’s own models and Grok and X benefits included. It also moves every third-party model onto prepaid credit at API list price. For a Cursor-models user, that is a good trade at a modest price. For a daily Claude user on legacy Pro, it takes a covered cost and turns it into a metered one.

Our advice by situation, as of October 30, 2026: take Premium if you are new to agents and want the full feature set; take Plus only if fixed-setting Cursor models are all you need; take Super or Ultra only when you can name the agents or Automations that will use the capacity; and if you are on legacy Pro and rely on third-party models, stay put until your own Spending tab says otherwise. Because Cursor describes the switch as permanent, the cost of waiting is low and the cost of a wrong move is high. Check Cursor’s pricing page before you pay, since prices are subject to change.

FAQ

Sources

Sources read October 30, 2026. Prices are quoted as Cursor states them on that date.

• Pricing and plans: Cursor help center (current plans, legacy mapping, optional and permanent switch, plan change steps, Auto and Router notes)
• Models & Pricing: Cursor docs (plan table, Plus fixed settings, per-model rates, top-ups and caps, Cursor Token Rate, Teams)
• Legacy Cursor plans: Cursor docs (legacy price table, two monthly pools, on-demand usage, what changes when you switch)
• Usage and limits: Cursor help center (weekly versus monthly pools, reset behavior, prepaid credit expiry, bring-your-own-key billing)
• Cursor pricing: cursor.com (plan prices, feature comparison, plan recommendations, reseller policy)

Related reading: Grok 4.5 and the Cursor launch, Cursor vs Antigravity, GitHub Copilot AI credits guide and Best AI model for coding, October 2026.