Cursor
A VS Code fork built around agentic editing, with its own Composer models, a pull request review bot and SOC 2 controls for teams that want AI in the editor.
Overview
It is a fork of VS Code, so extensions, keybindings and settings carry over on day one. What the fork adds is a tab model that predicts your next edit rather than your next token, an agent that plans across files and runs terminal commands, plus background agents that keep working while you do something else. Composer is Cursor's own model family, tuned for that specific loop rather than for chat.
For teams the relevant tier is Teams at $40 per user per month, which brings SOC 2 Type II, SSO, audit logs, admin policy enforcement and a privacy mode that stops code being stored on Cursor's servers. Bugbot is a separate purchase and does a different job: it reviews pull requests once they reach source control, where the in-editor agent review looks at local changes before they leave your machine. Enterprise adds pooled usage across seats, which is the only sane way to buy this if you have three people running agents constantly and twelve who barely touch them.
Two honest warnings. Credit accounting is the single biggest source of user complaint and the rules have changed several times in eighteen months, so budget with a real overage buffer rather than the headline $20. And on genuinely large repositories Cursor is slower than plain VS Code. Teams doing mostly terminal-driven work will get less out of it than teams who live inside an editor all day.
Key Features
- ✓ Composer 2.5, an in-house agent model shipped 18 May 2026, priced around $0.50 per million input tokens
- ✓ Tab model that predicts the next edit across a file rather than completing the current line
- ✓ Background agents that keep working on tasks outside the active editing session
- ✓ Bugbot, a pull request reviewer that runs after changes land in source control
- ✓ SOC 2 Type II, SSO, audit logs and a strict privacy mode that keeps code off Cursor's servers, from Teams upward
- ✓ Multi-model picker covering frontier models alongside the in-house Composer family
Where it holds
- • The tab and agent loop is still the reference experience, and on small to mid-sized repos it is not close
- • VS Code fork means near-zero migration cost for extensions, keybindings and settings
- • Composer 2.5 is fast and cheap enough that per-token model cost stops being the thing you optimise for
Where it breaks
- • Credit and quota rules have been rewritten repeatedly, most recently when Cursor Router replaced Auto mode on 22 July 2026 and split it into three tiers that meter differently
- • Max mode on a large refactor can consume a month of Pro credits in a single afternoon
- • Noticeably slower than stock VS Code on very large monorepos, because indexing and agent context cost something
- • Standardising a team on it means committing to one vendor's editor fork, which is a real lock-in question at org scale
My Take
Two model releases in two months tells you where the money is going: Composer 2 landed 19 March 2026 and Composer 2.5 on 18 May. The in-house models matter because they are cheap enough to run all day at roughly $0.50 per million input tokens, and because Composer 2 was built on a Kimi K2.5 base, which is a quietly interesting admission about where good open base models come from now. The editing loop itself is the best in the category. Billing is the weak point: Cursor Router replaced Auto mode on 22 July 2026 and split it into three tiers that meter very differently, and a Max mode refactor across a big repo can drain a month of Pro credits before lunch.
Quick Info
- Pricing:
- freemium
- Openness:
- Proprietary
- Starting at:
- Hobby free. Pro $20/month, Pro+ $60/month, Ultra $200/month. Teams $40 per user/month, Enterprise custom and annual only. Annual billing saves roughly 20%. Every plan includes a set amount of model usage with on-demand overage billed in arrears. Bugbot, the pull request reviewer, is sold separately at around $40 per user/month.
- Added:
- Aug 2026
- Updated:
- Aug 2026
Use Cases
Judge it on your own work
The notes above say where Cursor holds and where it breaks. The fastest check is your own workload.
Visit website ↗Cursor compared head to head
Alternatives to Cursor
GitHub Copilot
freemiumMicrosoft's AI coding assistant across VS Code, JetBrains and the GitHub web, with background agents that open pull requests and org-level policy controls.
Trae
freemiumByteDance's AI IDE and standalone SOLO agent that scaffolds projects, runs terminal commands and tests in a browser, at $10 a month for Pro against Cursor's $20.
Qoder
freemiumAlibaba's agentic IDE, which absorbed Tongyi Lingma (通义灵码) as Qoder CN on 20 May 2026. Repo Wiki and Quest Mode on credit-based pricing, with Linux builds since version 1.0.
Tencent CodeBuddy
freemiumTencent Cloud's coding assistant (腾讯云代码助手), a distinct product from the Hunyuan chatbot. Standalone IDE, editor plugins and a CLI with native Agent Client Protocol support.
Devin
freemiumCognition's autonomous coding agent, now bundled with the desktop IDE that used to be Windsurf, aimed at teams offloading well-specified tickets to background runs.