Cursor
A fork of VS Code rebuilt around agent-driven editing.
What it is
Cursor is what you get when the editor is designed for agents rather than having agents bolted on: multi-file edits, a diff you review before applying, and codebase indexing that actually understands the repo. The cost structure is the thing to read carefully — every tier is a usage pool, and third-party models are billed on top at list price.
Pricing updated 2026-11-11 Official pricing
Four times the Plus usage, cloud agents, code hosting, ad-free; most popular tier
Open pricingCentral billing, analytics, privacy controls; usage pooled 4–50× Plus
Open pricingCursor pricing explained
Every tier is a usage pool rather than a feature gate: $8 Plus, $30 Premium, $100 Super, $200 Ultra, with team seats at $30. The number on the page is not the number you will spend if you use frontier models heavily, because third-party model calls are metered separately at list price. That is not a hidden fee — it is the structure — but it does mean the monthly bill is a function of how much you delegate. Read the pool size before choosing a tier, and check the current page rather than a review, since Cursor has changed both names and limits more than once.
Free-tier limits are taken from the official pricing page on the update date above. They change often — confirm before you commit to an annual plan.
Cursor key features
Agent mode with reviewable diffs
Edits across many files at once, then shows the combined diff before you accept. The review step is what makes multi-file autonomy usable rather than alarming.
Codebase indexing
Builds an index of the repository so questions resolve against real symbols and call sites, not just the open file.
Tab completion with multi-line prediction
Predicts the next multi-line edit including the jump to the next location. Faster than typing, and the reason people describe it as harder to go back from.
Chat with file and symbol references
Point at specific files or symbols with @-mentions, which keeps the context small and the answer grounded.
VS Code compatibility
Extensions, themes and keybindings import directly. Migration cost is minutes, which is why the switch from VS Code is so common.
Cloud agents and automated review on higher tiers
Long-running tasks can run remotely, and reviews run on pull requests. These are the features that justify the upper tiers rather than the entry one.
Cursor: how it works
A fork, not a plugin
Because it owns the editor, it can act on the buffer directly — inline diffing, multi-cursor edits, jumping between files. A plugin is limited to the extension API and always feels one step removed.
Retrieval over an index of the repo
The index is what allows repo-wide questions. It is built locally and refreshed as files change; on a very large monorepo the initial build is the visible cost.
Bring-your-own model with metered third-party access
Frontier models can be used at list price, billed separately from the subscription. That is the honest structure behind the "usage pool" pricing, and the reason a heavy month can cost more than the sticker price.
Cursor core strengths
The most mature agent workflow of any editor
Multi-file edits, reviewable diffs and rollback are one continuous loop. Other tools have the same pieces, but the seams show more.
Low-risk entry at $8
Cheap enough to try on real work for a month. The decision to keep it is made on evidence rather than on a trial that expires mid-task.
Migration from VS Code is painless
Extensions and keybindings come across, so the cost of trying it is an afternoon rather than a week of setup.
Model choice at list price
No markup on third-party models means you can pick per task without wondering whether the vendor is taking a cut on the tokens.
Cursor compared with alternatives
| Tool | Cursor | GitHub Copilot | Cline |
|---|---|---|---|
| Best for | Individual developers who want the most mature agent-driven editor | Teams on GitHub that also want PR review and issue triage covered | Developers who want agentic editing with their own API key and no subscription |
| Pricing shape | Usage pools from $8 to $200 per month, third-party models billed separately | Free tier, then $10 / $39 / $100 per user per month | Extension is free; you pay the model provider directly per token |
| Deployment and data | SaaS with local codebase indexing; no self-hosting | SaaS; cloud agent runs in GitHub-managed environments | Runs locally in VS Code; can point at local models |
Written from the official product pages listed under Sources, not copied from another directory. Prices and limits reflect the update date above; features ship faster than reviews, so treat the list as a starting point and confirm on the official site.
Pros and cons
Strengths
- Agent workflow is the most mature of any editor here
- Cheap entry tier at $8 makes trying it low-risk
- Import of VS Code extensions and keybindings is painless
Limitations
- Third-party model usage is billed separately and is easy to underestimate
- Usage pools make monthly cost hard to predict for a heavy user
- You are committing to a separate editor rather than a plugin
Cursor use cases
Repo-wide refactors
Renaming a concept across dozens of files with a diff you can read in one pass, then accepting or rejecting by hunk.
Feature work from a spec
Give it the ticket and the relevant files; review the diff rather than describing each change.
Learning an unfamiliar codebase
Ask where a behaviour is implemented and follow the answer into real symbols instead of guessing from names.
Test generation alongside the change
Ask for the tests in the same session as the edit, so the assertions match what was just written.
Pairing on a design decision
Discuss two approaches with the actual code in context, then let the agent implement the one you pick.
