What Is Cursor? The AI Code Editor Explained

The name is the joke: a cursor is the blinking line where you type. Cursor is a code editor, built from the same open foundation as Visual Studio Code, that puts an AI agent at the centre of the writing experience.

Is Cursor a model?

No. Cursor is the product most often confused with the models inside it, so to be plain about it: Cursor does not make Claude or Gemini or GPT. It is a workbench that can use them, and you choose which company's model answers each request from a menu. The one model Cursor does train itself, Composer, is described further down.

What Cursor is made of

It also supports the wider toolkit of modern agents: MCP servers, reusable skills and hooks.

Its own model: Composer

Besides routing to other companies' models, Cursor trains its own coding model, Composer. Coverage of Composer 2 (March 2026) describes it as purpose-built for agentic coding and built on an open base model with extra training. The point of a house model is speed and cost: for routine steps of an agent loop, a fast specialised model can be cheaper than calling the most expensive frontier model every time.

Agent mode, in plain words

Older assistants suggested code and waited for you to paste it. Agent mode closes the loop:

1. You give a goal: "add a dark mode toggle and make the tests pass".
2. The agent searches your codebase to understand it.
3. It edits files, runs the build or tests, reads the errors and fixes them.
4. You review a diff and accept or reject.

You stay the reviewer. The agent does the legwork. Good habits: keep tasks small and specific, say how to verify success ("run the test suite"), and read the diff before accepting.

The Cursor CLI

The Cursor CLI brings the agent to the terminal. Install it with:

curl https://cursor.com/install -fsS | bash

In the interactive view you can use / for commands, @ to attach files and ! to run shell commands. It also has a headless mode for scripts and automation, including GitHub Actions, so you can run the same agent in your build pipeline. It works alongside whichever editor you prefer.

Is Cursor free?

Yes, there is a free tier, and no, it is not enough for daily work.

The Hobby plan costs nothing, needs no credit card, and gives you a limited number of agent requests plus access to Composer. It is genuinely useful for trying the editor and deciding whether the workflow suits you. What it will not do is carry you through a working week, because agent requests are the thing it limits and agent requests are the thing you actually want.

How much does Cursor cost?

Cursor is a subscription with included usage, not a pure per-token service and not a flat "unlimited" plan. From the pricing page:

The key to understanding the bill is the usage model. Each plan includes a set amount of model usage, and what that usage buys depends on which model you pick. A cheap, fast model stretches the allowance a long way; a top frontier model burns through it quickly. Cursor therefore sits between a flat subscription and an API bill: a prepaid allowance, with overflow billed afterwards.

If that distinction is unfamiliar, our guide to API versus subscription explains why the two are different products rather than two prices for the same thing.

Does Cursor train on your code?

This is the question that decides whether Cursor is usable at many companies, and it has a clear answer.

Cursor offers privacy mode, which an individual can switch on in settings and a team admin can enforce for everyone. Cursor's own wording is that when privacy mode is enabled, "we guarantee that code data is not used for training by us or our model providers."

Two things worth being precise about. That guarantee covers training, which is not the same as your code never leaving your machine: the agent still sends context to whichever model you chose in order to answer at all. And it is a setting, which means someone has to turn it on. If you are evaluating Cursor for work, turn privacy mode on first and confirm it with whoever owns that decision, rather than assuming the default.

Cursor vs GitHub Copilot

This is the comparison most people actually want, because Copilot is the incumbent and Cursor is the challenger.

The honest way to choose between them. Copilot is cheaper at the entry level and does not ask you to change editors, which matters more than people expect: your extensions, keybindings and team's setup all stay put. Cursor costs twice as much at the bottom tier and asks you to move, but the agent is built into the editor rather than bolted on, and that integration is the thing people who switch say they would not give up.

Price is not the deciding factor at these numbers. Both cost less than an hour of a developer's time per month. Decide on whether you want an editor with an agent inside it, or your current editor with an agent attached.

Cursor vs VS Code

Strictly, this is not a fair fight, because Cursor is VS Code, or rather it is built from the same open-source foundation. That fact carries practical consequences:

• Your extensions mostly work. Cursor can import your VS Code extensions, settings, themes and keybindings when you first run it, so the editor feels familiar within minutes.
• It is a separate application. You install Cursor instead of launching VS Code, and updates come from Cursor rather than Microsoft.
• Some Microsoft extensions are restricted. A handful of Microsoft's own extensions are licensed for Microsoft's products specifically, so they may not be available in a fork.

The straightforward version: if you want VS Code with AI, you can install Copilot into VS Code and keep everything. If you want an editor where the agent is the point rather than an addition, that is Cursor. Both are defensible; only one asks you to switch applications.

Cursor vs Claude Code and the other native agents

Neither is better in general. If you want one polished place, a visual diff review and the freedom to swap models, Cursor is attractive. If you live in the terminal, want your model maker's newest features the day they ship, or want to pay a single vendor directly, the native agents fit better. Many developers use both, and some even run Claude Code inside Cursor's editor.

Our comparison of the five coding agents goes through them one by one.

How to get started with Cursor

1. Download it from cursor.com for Windows, macOS or Linux.
2. Import your VS Code setup when it offers, so your extensions and keybindings come along.
3. Turn on privacy mode before you open anything you would not want used for training.
4. Start with tab completion for a day, before touching agent mode. It is the feature you will use most and it needs no learning.
5. Then give the agent one small, verifiable task, such as "add a test for this function and make it pass". Read the diff. That one exercise teaches you more about working with an agent than any tutorial.

What is Cursor AI primarily used for?

Cursor is used primarily for writing and changing code with an AI agent inside the editor. That is the whole product: everything else is in service of it.

In practice, the work divides into five jobs:

• Building a feature across many files in one go. You describe the outcome; the agent plans the change, edits every file it touches, and hands you a diff.
• Understanding an unfamiliar codebase. Ask "how does this work?" and get a guided tour with the relevant files opened, which is the use people underrate most.
• Fixing bugs. Paste an error, and the agent traces it, proposes a patch and runs the tests.
• Running several agents at once, on separate branches or in the cloud, so long tasks continue after you close the laptop.
• Automating review in CI, where Bugbot comments on pull requests or triages issues without anyone opening the editor.

What it is not primarily for: general chat, writing prose, or answering questions unrelated to your code. Cursor is a code editor first, and the models inside it are reached through a coding workflow rather than a blank chat box. If you want a general assistant, you want the Claude or ChatGPT app; if you want code changed in a repository, you want this.

Common questions

Links

• Pricing: cursor.com/pricing
• CLI: cursor.com/cli
• Comparison with other agents: CLIs and agent mode
• How AI billing works: API versus subscription