Our verdict · Coding agent
The strongest choice for developers who want AI inside the editor they think in: an agent window, Tab, cloud agents and Bugbot code review in one product. The plans changed shape after SpaceX agreed to buy Cursor in June 2026: the cheapest paid tier is now a Grok bundle, and third-party models are billed at API list price on top of the plan.
Pick it if
- Developers who want to stay in an editor and review every diff
- Teams that want AI code review on pull requests
- Mixed work: hand-editing, agent tasks and background agents
Skip it if
- You don't want to read code
- You want a fixed price for heavy use of Claude or GPT models
- You prefer working from a terminal or across many surfaces
Cursor is an AI-first code editor built on VS Code, with an agent window, Tab completions, cloud agents that work in the background, a CLI, and Bugbot code review on pull requests. In June 2026 SpaceX signed a binding agreement to buy Cursor's maker, Anysphere, for $60bn in stock (The Next Web), and the price list now reflects that: Cursor is bundled with xAI's Grok. This review compares Cursor's published plans and docs against our six criteria; we haven't benchmarked it.
Scores
- Speed to a first version3/5
Quick inside an existing project; for a brand-new app you still set up the stack and hosting yourself.
- Depth before the wall5/5
Built for large real codebases: rules, skills, subagents, hooks, cloud agents and review.
- Backend and data2/5
Brings no backend; it works well with whatever you run.
- Ownership and exit5/5
Your code stays in your repo and your infrastructure.
- Pricing predictability2/5
Usage pools are described as multiples, not numbers, and third-party model use is billed on top at API rates.
- Team and production readiness4/5
Teams with SSO and analytics, Bugbot reviews, and Enterprise SCIM and audit logs.
Overall 3.5/5, the plain average of six criteria. Scored by Mahmoud Halat from published specs and docs, checked Nov 1, 2026. How we score.
Pricing after the Grok bundle
| Plan | Price (USD) | What you get |
|---|---|---|
| Free | $0 | Limited agent and chat requests |
| Plus | $8/mo | Shared usage pool; Grok models only; Grok chat, voice and API access |
| Premium | $30/mo | 4x Plus usage, cloud agents, code hosting, AI code review, third-party models at API list price |
| Super | $100/mo | 15x Plus usage, priority model access at peak times |
| Ultra | $200/mo | 50x Plus usage, earliest access |
| Teams | From $30/user/mo | Central billing, usage analytics, SSO |
- Billed in
- A shared usage pool (Cursor plus Grok) refreshed weekly; third-party models billed separately at API list price
- Free tier
- Limited agent and chat requests
Source: cursor.com/pricing. Prices checked Nov 1, 2026, before tax. The page shows no annual prices and no request counts or dollar amounts for the usage pools.
Cursor's pricing page reads differently from a year ago, and the changes matter for coding:
- The $8 Plus plan is Grok-only. It includes Cursor plus Grok chat, voice, image and API access, but only Grok models in the editor. It's a good deal if Grok suits your work and a trap if you expected Claude or GPT.
- Premium at $30 is the real starting plan for most developers. It adds cloud agents, code hosting, AI code reviews and third-party models.
- Third-party models cost extra. The page says third-party model use is billed separately at API list price and is "not included in the monthly prices shown". Heavy use of Claude or GPT models in Cursor is pay-as-you-go on top of the plan.
- Usage is a shared, weekly pool. Plans are described as multiples of Plus (4x, 15x, 50x), shared across Cursor and Grok, refreshing weekly, with top-offs. No request counts or dollar values are published.
The result is a plan that's easy to buy and hard to forecast. If your team mostly uses one non-Grok model, compare Cursor's API-priced usage with a flat subscription like Claude Code's.
What it does well
Cursor is at its best for developers who want AI inside the editor they already think in. The agent works across files with planning, debugging and a design mode; Tab predicts edits as you type; cloud agents take tasks into the background, with options to run on self-hosted machines; and Bugbot reviews pull requests. It's customisable in the same terms as other modern agents, with rules, skills, MCP servers, plugins, subagents and hooks, and it connects to GitHub, GitLab, Azure DevOps, Bitbucket, Slack and Linear. Its own Composer models sit alongside Anthropic, Google, OpenAI and Grok models in the picker.
That combination suits mixed work: a developer hand-editing one file, an agent refactoring another, a cloud agent fixing a ticket, and Bugbot checking the result.
Where it struggles
Cursor is a developer tool. It doesn't host your app or provide a backend, and it assumes you can read and review code; people who want "describe it, get an app" should look at the builders instead. The other weakness is the business side: the plan structure changed shape this year, model access now depends on the Grok bundle, and the acquisition hasn't closed (SpaceX expected it in Q3 2026, subject to regulators). Teams buying annual commitments should expect the price list to keep moving.
Ownership
Nothing to export: Cursor works on your repository, on your machine or on cloud agents attached to your repo, and your code never lives inside Cursor. Switching editors costs you your rules and habits, not your work.
Key facts
- Made by
- Anysphere (SpaceX)
- Works in
- Its own editor (a VS Code fork), a CLI, cloud agents, Slack and Linear
- Models
- Cursor's Composer models and Grok, plus Anthropic, Google and OpenAI models at API price
- Backend
- None: it works on whatever stack your repo uses
- Ownership
- Your repository; nothing to export
Alternatives
- Claude CodeAgent-led work across terminal, IDE, desktop and cloud on a flat subscription
- Windsurf (Devin Desktop)The other agentic editor, now Devin Desktop
- GitHub CopilotCheapest seat, native to GitHub issues and PRs
Head-to-head: Cursor vs Claude Code · Cursor vs Windsurf
Questions people ask
How much does Cursor cost in November 2026?
Free has limited agent and chat requests. Plus is $8 a month but covers Grok models only. Premium, the first plan with cloud agents, AI code review and third-party models, is $30 a month; Super is $100 and Ultra $200. Teams starts at $30 per user a month. Checked November 1, 2026.
Why does Cursor's cheapest plan only include Grok?
Cursor's plans are now bundled with xAI's Grok products. SpaceX, which absorbed xAI, signed a binding agreement to buy Cursor's maker Anysphere in June 2026. The $8 Plus plan shares one usage pool between Cursor and Grok and is limited to Grok models; other vendors' models start at Premium.
Are Claude and GPT models included in Cursor's price?
Not in the plan price. Cursor's pricing page says third-party model usage is billed separately by usage, at API list price, and is "not included in the monthly prices shown". Premium and above give access to them.
Is Cursor better than Claude Code?
For people who want to stay in an editor and review every diff, usually yes. For agent-led work across terminal, IDE, desktop and cloud sessions on a flat subscription, Claude Code is the stronger fit. Our head-to-head covers the trade-offs.