Cursor’s pricing changed on October 12, 2026: new customers no longer buy a Cursor plan, they buy The X Subscription, a bundle of Cursor, Grok, Grok Bot and X Premium that starts at $8 a month. The headline price is lower than Cursor Pro ever was. The catch is in the footnotes, because Claude and GPT in Cursor are no longer part of any plan and are billed at API list price from the first request.
For a tool whose whole pitch was “pick the best model for the job,” that is a big shift. Here is what each tier includes, what the shared usage pool means for your coding time, and when switching makes sense.
What Is The X Subscription?
The X Subscription is one membership that covers Cursor, the Grok apps, Grok Bot and X Premium, with a single weekly usage pool shared across all of them. The xAI announcement says it launched for new customers on October 12 and would reach everyone globally over the following few days.
Engadget reports that the company behind it, now called SpaceXSI, acquired Cursor in August. This is the first time that ownership shows up in the product people pay for.
There are five tiers. Prices and usage multiples below come from the official pricing docs and Cursor’s pricing page:
| Tier | Monthly | Annual | Included usage | Third-party models in Cursor |
|---|---|---|---|---|
| Free | $0 | $0 | Limited | No |
| Plus | $8 | $80 | 1x (base) | No, Grok only |
| Premium | $30 | $300 | 4x | API list price, billed separately |
| Super | $100 | $1,000 | 15x | API list price, billed separately |
| Ultra | $200 | $2,000 | 50x | API list price, billed separately |
(Source: xAI, The X Subscription pricing docs and cursor.com/pricing, October 2026)
One correction to some early coverage: Engadget’s launch story said Ultra carries 20 times the Plus allowance. Both xAI’s docs and Cursor’s pricing page say 50x, so go with 50x.
The important Cursor features sit at Premium. Cursor’s pricing page lists cloud agents and automations, code hosting and AI code reviews, the latest Grok models and third-party model access on Premium, not Plus. Plus gets you Cursor with Grok models, the Grok apps, Grok API access and an X checkmark.
So the $8 entry price is real, but it is a Grok-only Cursor.
How the Shared Usage Pool Works
Every paid seat gets one weekly pool that Grok chat, Grok Imagine, Grok Bot and Cursor all draw from. The usage limits docs say the pool is shared across eligible Grok and Cursor surfaces rather than split into separate quotas, and that different activities draw at different rates.
This is the detail I would think hardest about before switching.
Under the old Cursor plans, your coding allowance was your coding allowance. Under The X Subscription, generating a video in Grok or letting Grok Bot run a task spends from the same weekly bucket as your Cursor agent. In the Cursor forum launch thread, one user who moved from Pro to Premium said exactly that: Grok Bot and Cursor agent usage now drain one quota, so total usage felt smaller. That is one person’s report, but it is the predictable outcome of a shared pool.
Three more rules from the docs matter for developers:
- The pool refreshes weekly, not monthly. A heavy Monday refactor cannot borrow from next week.
- Overage goes through Additional Usage credits. You buy top-ups, can enable Auto Top-up, and can set spending caps.
- The Grok API is separate. Paid plans include API access, but API calls are billed at standard model rates and do not come out of the pool.
On teams, self-serve seats each get their own pool. Seats do not share allowances, but the team shares one prepaid balance for Additional Usage.
Claude and GPT in Cursor Now Cost Extra
On The X Subscription, third-party models in Cursor are never included: they consume paid credits from the first request. The pricing docs state that third-party models “aren’t covered by included usage.” Cursor’s pricing page says the same thing more bluntly, listing “third-party models at API list price” as a Premium feature.
The vendor’s argument is that you get more for your money if you use Grok. The announcement claims each plan includes at least twice the coding usage with its own models as the comparable Cursor plan, and that Premium bundles SuperGrok, Cursor Pro and X Premium+, which it says previously cost $90 a month combined. Those are company claims, and neither the docs nor the pricing page publish token figures that would let you check the 2x.
Here is the honest tradeoff.
If you already use Grok models for most of your Cursor work, Premium at $30 looks like a better deal than Cursor Pro at $20 was, because you also get Grok Bot, the Grok apps and X Premium.
If you open Cursor to use Claude or GPT, the bundle does nothing for that spend. You pay $30 for the plan and then pay API list price for every Claude token on top. At that point you are paying Cursor a platform fee to use a model you could buy elsewhere at the same rate.
The forum thread is full of exactly this objection. The longest reply argues the new plans make Cursor less model-neutral, and asks for a purchasable plan that includes a third-party allowance. As of this writing, there is not one.
Should Existing Cursor Subscribers Switch?
No one is moved automatically, and for most Claude-heavy users the right move is to stay put for now. According to the docs, existing Cursor, SuperGrok and X Premium plans continue unless you change or cancel them. The switch option appears in your billing settings.
The rules that matter if you are on a legacy Cursor plan:
- You keep included third-party usage until you switch. The forum announcement says current Cursor subscribers keep their included third-party usage until they move. Switching is the point where Claude and GPT become pay-as-you-go.
- Refunds are separate from the new charge. Switching cancels eligible plans with a prorated refund to your original payment method, but the new plan is charged in full immediately.
- Apple billing is its own problem. If you pay for Cursor through the App Store, switching does not cancel it. You cancel in iOS settings and ask Apple for any refund.
- The old Ultra is closed. A Cursor staff reply in the forum says the previous Ultra plan is closed to new signups, and the new Ultra is a different plan.
My rule of thumb: open Cursor’s usage view and look at which models your last month actually went to. If Claude and GPT are most of it, switching turns included usage into a metered bill. If Grok already dominates, Premium is probably a better deal than what you have.
Cursor vs Claude Code vs GitHub Copilot on Price
For Claude-model coding, Cursor is now the most expensive way in, because the other two include Claude usage in the plan price. Here is how the entry tiers compare as of October 2026:
| Plan | Monthly price | What the price includes | Claude models included? |
|---|---|---|---|
| Cursor Plus (X Subscription) | $8 | Grok-only Cursor, Grok apps, X Premium features | No |
| Cursor Premium (X Subscription) | $30 | 4x usage pool, cloud agents, Grok Bot | No, API list price on top |
| Claude Pro | $20 | Claude apps and Claude Code | Yes, within plan limits |
| GitHub Copilot Pro | $10 | $15 in AI credits, unlimited completions | Yes, billed from credits |
(Sources: docs.x.ai, claude.com/pricing, GitHub Copilot plans, October 2026)
Claude Pro is $20 a month, or $17 a month billed annually, and Anthropic’s pricing page says Claude Code is included in all paid plans. GitHub Copilot Pro is $10 a month with $15 in AI credits after GitHub added flex allotments, as covered in the GitHub Copilot pricing breakdown.
The chart below shows the part xAI’s pricing table does not spell out: what you pay per Plus-sized slice of included usage at each tier.
- Plus ($8, 1x) $8.00
- Premium ($30, 4x) $7.50
- Super ($100, 15x) $6.67
- Ultra ($200, 50x) $4.00
Source: The Vibelog calculation from xAI, The X Subscription pricing docs (Oct 2026)
Premium is barely cheaper per unit than Plus. The real volume discount only arrives at Ultra, where usage is half the per-unit price of Plus. That fits a bundle designed to pull heavy users up the ladder, and it matters less if your heavy usage is Claude, because none of it touches the pool.
For a broader feature comparison beyond price, the Claude Code vs Cursor vs GitHub Copilot comparison is still the best place to start. Price aside, Cursor’s editor, agents and review features did not get worse this week.
Cursor for Teams After the Change
Team seats cost the same as individual tiers: $30 for Premium, $100 for Super and $200 for Ultra per seat, and Plus is not offered to teams. The docs say a team can mix seat tiers, for example three Premium seats and one Super.
The announcement claims this is cheaper per seat than the old Cursor Teams plan with three times the coding usage. Treat that like the 2x claim: plausible for Grok-heavy teams, irrelevant for teams standardised on Claude or GPT.
Team plans include SSO, admin controls, usage analytics and consolidated billing. Third-party model spend for the whole team comes out of a shared prepaid balance, so the spending cap becomes the setting finance will care about.
If your team is evaluating alternatives, the subscription market for builders looks very different by price point. App builders like Lovable and Bolt.new still sell $25 Pro plans, as covered in the Lovable vs Bolt.new vs v0 comparison, but they are aimed at a different job than an AI editor.
The Verdict on Cursor’s New Pricing
The X Subscription is a good deal if you are happy coding with Grok models and you would use Grok or X Premium anyway. Plus at $8 is the cheapest paid AI editor entry I know of, and Ultra’s 50x pool is a strong rate for very heavy Grok usage.
It is a worse deal for the developers who made Cursor popular: people who paid for an editor and picked Claude or GPT inside it. For them, the new plans turn included usage into a metered bill on top of a higher base price.
If that is you, do not rush the switch. Your existing plan keeps its third-party allowance until you move, and nothing in the docs forces you to move yet.
