Cursor

Where Cursor sits now

Cursor is the AI code editor that made the whole category compete on context: built on VS Code, it indexes your repository and lets an agent read, question and edit across many files at once. That product is still here — the editor, the agent, the tab completions. What changed is who sells it and how you pay for it. Cursor is now part of The X Subscription, the xAI plan that bundles Grok, Grok Bot, Cursor and X Premium together (x.ai/news/x-subscription). Third-party reports put the corporate change at a SpaceX acquisition of Anysphere in June 2026 at a reported $60B valuation, with Cursor sitting under the xAI umbrella; xAI's own legal documents reference agreements with Anysphere and a "SpaceXSI" entity, so the corporate shape is new and worth understanding before you sign anything.

For new customers this means the standalone Cursor plans are legacy. At time of writing the X Subscription runs Free ($0), Plus ($8/month), Premium ($30/month), Super ($100/month) and Ultra ($200/month), with Teams seats at $30, $100 or $200 per user per month matching those tiers (cursor.com/help/account-and-billing/pricing). Existing Hobby/Pro ($20)/Pro+ ($60)/Ultra ($200) and Teams Standard ($40)/Teams Premium ($120) subscribers keep their plans, prices and features and can stay — switching is optional and permanent.

What it actually does

The editor covers the full loop of AI-assisted development. Tab completions predict your next edit, not just your next line. The agent window reads your codebase and makes multi-file changes from a prompt. Cloud agents work on tasks outside your local machine, Automations handle recurring jobs, Bugbot reviews pull requests, and there is a CLI and an iOS app — the Cursor Router now spans the Agents window, editor, CLI, Cursor SDK and iOS. MCP servers and skills extend what the agent can reach.

The model lineup is now Grok-centric: paid plans include a weekly usage pool covering Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5, shared across Cursor and the Grok apps. The plan matrix gates features hard. Plus runs the included models at fixed settings — no Auto, no Cloud Agents, no Automations, no Bugbot, no Origin, no third-party models. Premium and above add all of those, with third-party models billed at API list prices from prepaid credit. Auto mode itself split into Cost, Balance and Intelligence: individual plans get Cost mode only, while Teams and Enterprise get Cursor Router with Balance and Intelligence routing (routed third-party usage bills at that model's list price plus a token rate).

My testing notes

The context engine remains the reason to pick this editor. In my testing, agent tasks on a mid-sized repository with scattered config and duplicated helpers came back with edits in the right files far more often than editors that treat each open file as an island — a refactor touching a route, its tests and a type definition was proposed as one coherent change. Tab completions on repetitive boilerplate saved real keystrokes, and when a suggestion was wrong, the next-edit prediction style made the correction path shorter than regenerating a whole block.

Second, the plan gates are the thing that surprised me in daily use. On Plus I hit the boundary of "fixed settings" quickly: no Cloud Agents and no Auto meant every non-trivial task was a local agent run at medium effort. For me the practical floor for professional work is Premium at $30/month — which is also the plan xAI markets as replacing three separate subscriptions (SuperGrok, Cursor Pro and X Premium+, "previously $90/month combined" per the announcement). If you were already paying for those, the bundle math is genuinely good; if you only wanted the editor, you are now paying for a bundle whether you use Grok or not.

Third, the migration mechanics deserve care. Billing for new plans runs through grok.com/dashboard/billing, not the old Cursor dashboard; legacy plans still bill through the Cursor dashboard and Stripe. Switching from a legacy plan cancels it with a prorated refund and cannot be undone. In my reading of the docs, legacy usage pools (Cursor Models and Other Models) and the new weekly shared pool are different meters, so comparing "what I get for $20 now vs $30 next" takes more arithmetic than it used to.

For teams, the enterprise tier adds the controls that procurement asks for: pooled usage across Grok and Cursor, SCIM provisioning, audit logs, advanced admin controls and invoice billing, while Teams seats carry SSO, consolidated billing and usage analytics according to the X Subscription announcement. In my testing of the team-facing settings, the usage analytics are the part I would show a finance lead: because everything now draws from a shared weekly pool, per-seat consumption is visible in one place instead of scattered across two legacy meters. Whether your organisation is comfortable with a single AI vendor sitting under both your chat assistant and your editor is a governance question more than a technical one — it concentrates negotiating power in one contract.

Against GitHub Copilot, the current contrast is sharper than it was. Cursor's agent works against a locally indexed codebase and edits across files inside the editor; Copilot's coding agent works through GitHub itself, opening pull requests from issues. I find Cursor's in-editor loop faster for building, and Copilot's repository-native flow better for delegated background tasks — and many teams I talk to now run both, which the per-seat pricing on each side makes affordable.

Where it falls short

Privacy remains the standing question for proprietary code. Codebase indexing and agent runs mean your code goes to Cursor's servers unless you configure otherwise; teams with strict IP rules need to review privacy mode and enterprise controls before adoption — a point that was true before the acquisition and is more pointed now that the operator is a different company.

The product surface is also fragmenting. Three Auto modes, two billing systems, legacy and X-Subscription plan ladders, model pools that differ by tier — the editor is still excellent, but the commercial layer around it is more complex than it was in 2025. Plus is a teaser tier; real work needs Premium and up. The Grok-weighted model lineup is good news for Grok fans and a change for teams standardised on other providers, who now pay third-party models at list price from prepaid credit. And acquisitions of this scale always bring roadmap risk: prices, included usage and feature gates have all moved once this year and may move again.

One last practical point on models: because the weekly pool is shared across Grok and Cursor, heavy chat use in the Grok apps draws from the same allowance as your coding agents. In my testing it paid to think of the pool as one budget for AI work across both products rather than an editor allowance with chat as a bonus.

Verdict

I rate Cursor 4.2. On the rubric it is still the deepest codebase-context experience in an editor I have used, and the agent/tooling surface (Cloud Agents, Bugbot, CLI, SDK, iOS) is broad; usability stays high because it is VS Code underneath. Value is real on the bundle if you use Grok or X, confusing if you just want an editor, and the reliability score carries the acquisition and billing-migration churn. If you are a professional developer whose work lives in a codebase larger than a few files, Cursor on Premium or above is the setup I would pick today — with the privacy conversation had first for proprietary code.

Sources

• https://cursor.com/help/account-and-billing/pricing
• https://x.ai/news/x-subscription
• https://docs.x.ai/x-subscription/overview
• https://x.ai/legal/x-subscription-cursor-addendum
• https://x.ai/legal/x-subscription-grok-addendum
• https://cursor.com/docs/models-and-pricing
• https://cursor.com/help/the-x-subscription/overview
• https://siliconvalleyinvestclub.com/ko/companies/cursor/
• https://platform.tracxn.com/a/d/company/560112cfe4b04ab30a97ee2b/cursor
• https://vibecoding.app/blog/cursor-pricing
• https://continuumcode.ai/guides/cursor-pricing