Cursor is a polished AI first code editor with agents, cloud agents and a wide choice of frontier models, free to start and $20 a month for Pro. It suits developers who want AI built into the editor itself, but usage limits are not published as numbers, so heavy users can find costs hard to predict. Claude Code gives most developers more agent power for the same $20.
What Cursor is
Cursor is an AI coding product from Anysphere. It started as a code editor with AI built in, and it now also runs in the terminal as a command line tool, as a web agent, in Slack and in GitHub, where it reviews pull requests Cursor ↗.
The core idea is that you hand tasks to an agent instead of typing every line. Cloud agents go further: they run on their own machines, in parallel, for hours or days, and build, test and demo features for you to review Cursor ↗.
You can choose between models from OpenAI, Anthropic, Google, SpaceXAI and Cursor's own Composer models Cursor ↗. Cursor says it is trusted by over half of the Fortune 500 Cursor ↗. Hands on testing is in progress, and we will update this review with our results.
What it does well
Model choice is wide. Paid plans include frontier models, and the home page lists options such as Opus 5.5, GPT-5.6 Sol, Gemini 3.1 Pro, Grok 4.7 and Composer 2.5 Cursor pricing ↗ Cursor ↗. You are not tied to one AI vendor.
Autocomplete does not run out on paid plans. Pro, Pro Plus and Ultra include unlimited Tab completions Cursor docs ↗.
Cloud agents and code review are built in. Pro includes cloud agents and Bugbot, Cursor's code review bot Cursor docs ↗. Teams adds agentic code reviews with Bugbot, cloud agents and automations with shared team context, and a team marketplace for internal rules, skills and plugins Cursor pricing ↗.
It works with the wider agent tool ecosystem. Paid plans support MCPs, skills and hooks Cursor pricing ↗.
Teams get real controls. Teams includes centralized billing, usage analytics, team wide privacy mode and SAML or OIDC single sign on Cursor pricing ↗. Enterprise adds pooled usage, SCIM seat management, audit logs and model and MCP access controls Cursor pricing ↗. Cursor lists SOC 2, ISO 27001 and ISO 42001 certifications Cursor ↗.
On demand usage keeps quality steady. When you pass your included usage you can keep going at API rates, and Cursor says requests are never downgraded in quality or speed Cursor docs ↗.
Where it falls short
Limits are vague. The pricing page says every plan includes a set amount of model usage, but it does not publish numbers Cursor pricing ↗. Its docs describe extended agent limits without a dollar figure Cursor docs ↗. That makes monthly costs hard to forecast if you lean on the agent.
Overages are billed after the fact. On demand usage is billed in arrears at API rates Cursor pricing ↗ Cursor docs ↗, so a busy month can cost more than the plan price.
Adoption is slipping. In JetBrains' 2026 survey of professional developers, Cursor's work adoption fell from 18% to 12%, while Claude Code rose to about 39% JetBrains survey ↗.
The free plan is thin. Hobby offers limited agent requests and access to Composer Cursor pricing ↗, which is enough to try the editor but not to work in it every day.
We did not find a yearly discount on the pricing pages we checked Cursor pricing ↗ Cursor docs ↗.
Pricing and plans
Hobby is free with no credit card. It includes limited agent requests and access to Composer Cursor pricing ↗.
Pro is $20 a month with extended agent limits, frontier models, cloud agents, MCPs, skills and hooks, and Bugbot Cursor pricing ↗ Cursor docs ↗. Pro Plus is $60 a month and Ultra is $200 a month, with the same features and more usage Cursor docs ↗. Cursor recommends Pro Plus for daily agent users and Ultra for agent power users Cursor pricing ↗.
Teams Standard is $40 per user a month, and Teams Premium is $120 per user a month with five times the Standard agent limits Cursor docs ↗. Enterprise is custom priced and adds invoice billing and admin controls Cursor pricing ↗.
Usage beyond your plan is billed monthly at API rates Cursor docs ↗. Prices exclude taxes, and self serve plans accept major credit and debit cards Cursor pricing ↗. There is no free trial of paid plans listed.
Who should (and shouldn't) use it
- Good fit: developers who want AI built into a full editor and like switching between models.
- Good fit: teams that want shared rules, privacy mode, SSO and code review in one product.
- Skip it if: you want predictable costs with published limits.
- Skip it if: you are not a coder and want an app built for you. An app builder fits better.
- Skip it if: you mainly want long running agent work from the terminal. Claude Code is our pick for that.
Alternatives worth a look
- Claude Code: our top pick, an agent that works across your whole project from the terminal, IDEs, desktop or web Claude Code ↗, included with Claude Pro Claude pricing ↗.
- GitHub Copilot Pro: $10 a month with unlimited inline suggestions and $15 in monthly AI credits Copilot plans ↗.
- Devin Desktop: the editor formerly called Windsurf, with a free plan that includes unlimited Tab completions Devin pricing ↗.
- OpenAI Codex: included in every ChatGPT plan, with the CLI and IDE extension on Plus at $20 a month Codex pricing ↗.
Where it appears
ComparedThe Best AI Coding AssistantWhat’s good
- Polished AI editor
- Cloud agents
- Team privacy mode
What to know
- No numeric limits published
- Falling adoption
Key features
- Agent with extended limits on Pro
- Frontier models
- Cloud agents
- MCPs, skills and hooks
- Bugbot code review
- Teams privacy mode and SSO
Questions people ask
Is Cursor free?
Yes. The Hobby plan is free with no credit card, but it only includes limited agent requests Cursor pricing ↗.
How much does Cursor cost?
Pro is $20 a month, Pro Plus $60 and Ultra $200. Teams starts at $40 per user a month Cursor docs ↗.
What happens when I run out of Cursor usage?
You can keep working with on demand usage, billed monthly at API rates, without a drop in quality or speed Cursor docs ↗.
Is Cursor better than Claude Code?
They work differently. Cursor is an editor first, while Claude Code is an agent you steer from many places Cursor ↗ Claude Code ↗. More developers now use Claude Code at work, about 39% versus 12% for Cursor in JetBrains' data JetBrains survey ↗.
Where this comes from
Checked Oct 9, 2026. Vendor pages describe their own products; independent sources add outside views. Hands on testing is in progress.
01