Tool profile
Cursor: what it is, what it costs, where it breaks
Cursor is an AI-native code editor made by Anysphere that combines tab completion, chat and an autonomous agent in one application. Developers use it to write and change code across multiple files, and it also offers cloud agents and automated pull request review on paid plans.
What to know first
Cursor asks you to adopt its own editor, and in return the completions, chat panel and agent all share the same project context. Usage on paid plans is metered in two pools, one for Cursor's own models and one for third-party models billed at each model's API price. Privacy Mode is the setting to check first, because it decides whether your code can be used for training.
Cursor at a glance
| Maker | Anysphere, United States |
|---|---|
| Free plan | Yes |
| Entry price | $20 per month (Pro) |
| API access | Yes |
| Category | Best AI Coding Assistants in 2026, rank 1 of 9 |
| Website | cursor.com |
Who Cursor is best for
Full-stack developers
Work that spans frontend, backend and config files benefits from an agent that edits several files in one pass and shows a reviewable diff before anything is accepted.
Developers who want to choose models
Cursor exposes its own models alongside third-party ones and an Auto mode, so you can trade cost against capability per task instead of living with one model.
Small engineering teams
Teams plans add centralized billing, team-wide privacy mode, usage analytics and SSO, which makes a group rollout manageable without a full enterprise contract.
Cursor plans and pricing
| Plan | Price | What it includes |
|---|---|---|
| Hobby | Free | No credit card required. Limited agent requests and access to Cursor's Composer model, enough to evaluate the editor. |
| Pro | $20 per month | Unlimited tab completions, extended agent limits, cloud agents and Bugbot, with included usage in two model pools. |
| Pro Plus | $60 per month | Same feature set as Pro with a larger included usage allowance for heavier daily agent work. |
| Ultra | $200 per month | Highest individual usage allowance, intended for people who run agents for much of the working day. |
| Teams | $40 per user per month | Standard tier adds centralized billing, shared cloud agents, usage analytics, team-wide privacy mode and SAML or OIDC single sign-on. |
Prices as of October 2026; vendors change plans often, so confirm on the vendor's pricing page.
What Cursor does
- Agent mode
- The agent plans a change, edits files, runs terminal commands and iterates on errors. Edits appear as diffs you accept or reject, so the human stays the reviewer for each change.
- Tab completion
- A completion model predicts the next edit, not only the next token, and proposes multi-line changes as you type. Paid plans list tab completions as unlimited.
- Cloud agents
- Agents can run remotely on Cursor's infrastructure while you keep working, and Teams plans share context across them. This is the closest Cursor gets to a hand-off workflow.
- Bugbot review
- Bugbot reviews pull requests for defects before a human reads them. It is billed on usage on individual plans and appears as a team feature on the Teams tier.
- MCP, skills and hooks
- Cursor supports Model Context Protocol servers, skills and hooks, so the agent can reach issue trackers, databases and internal tools that you connect to it yourself.
How Cursor handles your data
Cursor says that with Privacy Mode on, customer data is not used for training by Cursor, and that it has zero-retention agreements with its model providers. Prompts that trigger abuse detectors may be kept for investigation. With Privacy Mode off, Cursor may store and use codebase data, prompts, editor actions and snippets to improve its features. Privacy Mode is available on free and Pro accounts, and Teams has a team-wide setting. Cursor states a SOC 2 Type II attestation and publishes subprocessors on its trust portal. Retention periods are not stated on the pages checked.
Where Cursor falls short in practice
- Usage runs down at different speeds. Included usage is split into a Cursor-models pool and an other-models pool. Picking an expensive third-party model for every task drains the second pool faster, and extra usage is billed on demand.
- You must move editors. Cursor is its own application. Developers who depend on JetBrains refactoring tools or Visual Studio debugging would be trading away a working setup, not adding a plug-in.
- Some pricing is not published. The pricing page does not give dollar amounts for included usage, and the Teams Premium tier price is not shown. Budgeting for a team needs a sales conversation or a pilot month.
Works with
Alternatives to Cursor
Cursor: common questions
Does Cursor train on my code if I turn Privacy Mode on?
No. Cursor states that with Privacy Mode enabled, customer data is not used for training by Cursor, and that its model providers do not retain or train on the data under zero-retention agreements. Prompts that trigger abuse detectors may be kept for investigation and then deleted under the provider's policy.
How much does Cursor cost for one developer?
As of October 2026, Cursor Pro is $20 per month, Pro Plus is $60 and Ultra is $200. Hobby is free with limited agent requests. The higher tiers raise how much model usage is included before on-demand charges apply, which is the practical difference between them.
Can I use my own model API key in Cursor?
Cursor's data page says that using your own API key does not bypass its backend, which still handles final prompt construction. So a personal key changes who pays the model provider, not whether Cursor's servers see the request. Check the current docs for which features accept a key.
What is the difference between Cursor's two usage pools?
One pool covers Cursor's own models, which get significantly more included usage. The other covers third-party models and is consumed at each model's API price. Auto routing bills each request at the list price of whichever model it picks, so model choice drives how long your allowance lasts.
Is Cursor SOC 2 certified?
Cursor's security page states it holds a SOC 2 Type II attestation, along with AIUC-1 and ISO certifications, with reports available on request through its trust portal. The page does not name the ISO standard, so ask for the report if your procurement process requires the exact scope.