What Is Cursor and Who Is It For? (2026)

Cursor is a code editor, not a plugin: a fork of VS Code with AI wired into the editing loop. It does three things an assistant bolted onto an editor cannot do as well: completions that read the surrounding file, a chat that has indexed your whole repository rather than the open tab, and an agent that plans and applies edits across many files at once. Two things have changed about it this year that most explainers have not caught up with. Its parent company was acquired by SpaceXAI in August 2026, and its pricing is now a weekly usage budget rather than a monthly seat.

This guide gives the definition, then the ownership change, then how it genuinely differs from GitHub Copilot, what it costs, and who should not switch. For the wider field see our coding tool reviews and our picks for developers.

What does Cursor actually do that an extension cannot?

Three things, and they all come from owning the editor rather than living inside it.

• Codebase-aware chat. Cursor indexes your repository, so you can ask where something is handled or why a function exists and get an answer grounded in files you never opened. An assistant that sees only your open tabs cannot do this as well, and on a large unfamiliar codebase it is the single most useful feature here.
• A multi-file agent. You describe a change, it proposes edits across every file involved, and you review a diff. Renaming a concept across thirty files, or threading a new parameter through a call chain, is exactly the kind of tedious correct work this does well.
• Completions with real context. The inline suggestions read the surrounding code rather than just the line above, which makes them noticeably better at matching your conventions.

What it does not do is remove the need to read the output. The agent is confidently wrong often enough that reviewing its diff is the job, not a formality, and anyone treating it as a vending machine will ship bugs faster than before. That is true of every tool in this category and it is worth saying plainly rather than discovering it in production.

Who owns Cursor now, and does it matter?

Anysphere, the company behind Cursor, was acquired by SpaceXAI in a $60 billion all-stock deal. Reporting traces it to an option announced in April 2026 and a close in mid-August, with sources differing by a day on the exact date, after which Cursor became a wholly owned subsidiary. The visible consequence inside the product is Grok: Grok models are available in the editor, and Grok products draw on the same usage pool as your Cursor plan.

Whether that changes your decision is a question about the company rather than the software, and reasonable teams land on both sides of it. The concrete things worth weighing:

• Model concentration. A single owner behind the editor and a frontier model family is convenient and also a dependency. If you want model choice, note that third-party models in Cursor are billed separately at API list price rather than from your pool.
• Procurement and policy. Some organisations maintain approved-vendor lists, and an ownership change can move a tool on or off one. Worth checking before a team rollout rather than after.
• Nothing about the editor got worse. The product has kept shipping, and the acquisition is not a reason to avoid it on quality grounds.

How is it different from GitHub Copilot?

Structurally, and the structure is the decision. GitHub Copilot is an assistant you add to the editor you already use, so your setup stays exactly as it is. Cursor is the editor, so you move into it and the AI gets more to work with in return. Everything else follows from that, which is why feature-by-feature comparisons of the two age so badly: the lists converge every quarter and the shapes never do.

The billing has converged too, interestingly, and both now meter the same way. Copilot includes unlimited inline completions on paid plans and explicitly does not charge credits for them, while agent mode, code review and CLI work draw on a dollar-denominated credit allowance. Cursor pools everything into one cost-measured allowance instead. So on Copilot your autocomplete is free and your agent costs money; on Cursor both come out of the same bucket.

We go through autocomplete quality, agent behaviour and team fit properly in our head-to-head comparison, which is the place to go if you have already decided you are willing to switch editors and just want to know which is better.

What does it cost, and which tier should you start on?

Free, then Plus at $8, Premium at $30, Super at $100 and Ultra at $200, with Teams from $30 per user and pooled usage. The headline prices are the easy part. The part worth understanding is that you are buying a budget, not a quota: each tier includes a pool of AI usage measured by cost rather than by request count, shared across Cursor and Grok, and it refreshes weekly. Cursor's own FAQ puts the overage choice plainly: if you reach your included usage, "you can keep going with top-offs, or wait for your allowance to refresh".

Two practical consequences. A weekly refresh is better than a monthly one for recovering from a heavy day and worse for a sprint that runs hot for a fortnight, so your usage shape matters more than your average. And because the pool is denominated in cost, an expensive model burns it faster than a cheap one for the same number of requests, which makes model choice a budgeting decision rather than only a quality one.

Start on Plus rather than Premium. Most individuals who think they need $30 a month have not yet learned which tasks are worth handing to the agent, and the cheapest way to learn that is on a small budget that runs out and teaches you something.

Who should stay on VS Code?

• Anyone with a deeply customised setup. Cursor is a VS Code fork so most things carry over, but if your environment is a decade of accumulated configuration, the migration tax is real and Copilot costs you none of it.
• Developers whose employer mandates an editor or a vendor. Not a judgement about quality, just a constraint, and Copilot sits inside what is already approved at most GitHub shops.
• People who mainly want better autocomplete. That is the part where the two are closest, and on Copilot paid plans completions do not draw down a credit budget at all.
• Teams who need predictable monthly cost. A cost-denominated pool that refreshes weekly is harder to forecast than a flat seat price, and finance departments notice.
• Anyone whose organisation has a view on the owner. Check the approved-vendor list before a rollout rather than after.

So who is Cursor for?

Developers who spend most of the day in an editor, work on codebases large enough that finding things is a real cost, and are willing to change tools to get a better agent. For that person it is the strongest version of this idea available, and $8 a month to find out is close to free. We rate it 4.7 out of 5, a shade above Copilot, and the gap is almost entirely the agent and the codebase indexing rather than anything about autocomplete.

If you are not that person, Copilot is not a consolation prize: it is the correct answer for anyone who should not be changing editors this quarter. Read the full scoring in our Cursor review, compare the assistant approach in our GitHub Copilot review, or install the free tier and give it one real repository for a week before you decide.

Frequently asked questions

What is Cursor?+

Cursor is a code editor built as a fork of VS Code, with AI wired into the editing loop rather than added as an extension. Three things make it distinct: inline completions that read the surrounding file, a chat that has indexed your whole codebase rather than just the open tab, and an agent that can plan and apply edits across many files at once. Because it is a VS Code fork, your extensions, keybindings and themes mostly carry over, which is why switching feels smaller than it sounds.

Who owns Cursor?+

SpaceXAI. Anysphere, the company behind Cursor, was acquired in a $60 billion all-stock deal that reporting puts at closing in mid-August 2026, following an option announced that April. Cursor now operates as a wholly owned subsidiary, and the practical sign of it is Grok models appearing inside the editor and Grok products sharing the same usage pool as your Cursor plan. Whether that matters is a judgement about vendor concentration and procurement rather than about the editor, and plenty of teams will reasonably land on either side.

How much does Cursor cost?+

There is a free tier with limited AI usage, then Plus at $8 a month, Premium at $30, Super at $100 and Ultra at $200, with Teams from $30 per user. The thing to understand is that you are not buying a number of requests: each tier includes a pool of usage measured by cost and shared across Cursor and Grok, and it refreshes weekly rather than monthly. Cursor’s own wording is that if you reach your included usage "you can keep going with top-offs, or wait for your allowance to refresh".

Is Cursor better than GitHub Copilot?+

They answer different questions, so the honest answer depends on whether you want to change editors. Copilot is an assistant you add to the tools you already use, and on a paid plan its inline completions cost nothing against your credits. Cursor is a whole editor you move into, and in exchange the AI sees more context and its agent is better at multi-file work. If your current setup is deeply customised or mandated by your employer, Copilot wins by default. If you are willing to switch, Cursor is the stronger agent experience.

Is Cursor free?+

There is a genuine free tier, and it is a trial of the idea rather than a working setup. It gives limited agent and chat requests, excludes cloud agents and third-party models, and will run out quickly on any real task. That is enough to answer the only question that matters at the start, which is whether the editor feels better than yours. Use it for a week on a real repository, and if the agent earns its place, Plus at $8 is where most individuals should land first rather than jumping to Premium.