Cursor
An AI code editor with an agent, Tab completion, cloud agents and code review, offering models from several vendors.
- Free plan
- Free trial
- Agentic tool
- Vendor-sourced
What is Cursor?
Cursor is an AI-native code editor from Anysphere that combines an in-editor Agent, Tab autocomplete and cloud agents, with access to models from OpenAI, Anthropic and Google alongside Cursor's own models.
- Best forDevelopers who prefer an editor-centred workflow
- PricingFree Hobby plan; Pro $20 per month. Free plan available.
- Main limitationUsage is metered at model API prices against an included amount, so heavy agent users may incur on-demand charges.
- AlternativesClaude Code
Cursor overview: who it is for and what it does.
Cursor is a code editor built around AI assistance. Its pricing page lists an Agent that developers can use for coding tasks, Tab completions powered by Cursor's own models, cloud agents, and Bugbot for agentic code review. Paid plans add extended Agent limits, access to frontier models, and support for MCP servers, skills and hooks.
Cursor describes itself as powered by a combination of custom models and models from other vendors. Its pricing documentation lists usage in terms of model API prices: individual plans include an amount of agent usage, and on-demand usage can continue after the included amount is consumed, billed in arrears.
For organisations, Teams adds shared chats, commands and rules, centralised billing and administration, org-wide Privacy Mode controls and SAML or OIDC single sign-on. Enterprise is custom-priced for organisations that need invoicing, pooled usage or advanced security.
Main problem it solves
Helps developers write, edit and review code faster by putting an AI agent, autocomplete and code review inside the editor where they already work.
Typical customer segment
Individuals through enterprises, with Hobby, Pro, Pro+, Ultra, Teams and Enterprise plans.
Target users
- Individual developers who want an AI-native editor
- Engineering teams that want shared rules and org-wide privacy controls
- Organisations needing invoicing, pooled usage and advanced security on Enterprise
Primary use cases
- Multi-file feature work with an in-editor agent
- Autocomplete with Tab
- Agentic code review with Bugbot
- Running agents in the cloud
- Connecting tools through MCP, skills and hooks
What makes it distinct
- Models from several vendors in one product, plus Cursor's own models for features such as Tab.
- An editor-centred workflow with the agent, completions and review in one place.
- Org-wide Privacy Mode enforcement and SSO on team plans.
Cursor features explained.
Coding assistance
- Agent
- An in-editor agent for coding tasks. The free Hobby plan includes limited Agent requests; paid plans extend the limits.
- Tab completion
- Cursor's native autocomplete, powered entirely by Cursor's custom models according to its pricing FAQ.
- Max Mode
- Certain models can use Max Mode for longer reasoning and larger context windows, up to 1M tokens according to Cursor's documentation.
Agents and review
- Cloud agents
- Agents that run remotely. Cursor's documentation says cloud agents are charged at model API pricing.
- Bugbot
- Agentic code review. Cursor lists Bugbot as available on usage-based billing for individuals and included for agentic code reviews on Teams.
Extensibility
- MCP servers, skills and hooks
- Listed on Pro and above for connecting external tools and customising agent behaviour.
- Shared rules, commands and chats
- Teams adds shared chats, commands and rules, and a team marketplace for internal rules, skills and plugins on some plan versions.
Administration
- Org-wide Privacy Mode controls
- Teams admins can enforce Privacy Mode across the organisation.
- SSO and admin dashboard
- Cursor's documentation lists SAML or OIDC SSO, an admin dashboard with usage stats and centralised billing for team plans.
Cursor pricing and plans.
Prices from a vendor-published source. Checked 15 Oct 2026 against Cursor (cursor.com) · Cursor (cursor.com/docs) · Cursor (cursor.com).
Pricing model
Monthly subscription tiers with included usage priced at model API rates; on-demand usage billed in arrears after included usage is used.
Free plan
Yes
Free trial
Yes Cursor's pricing page mentions a Pro trial for new accounts; the length has differed between versions of the page.
Cursor plans and prices (USD)
Hobby
Free
n/a
No credit card required. Limited Agent requests and limited Tab completions.
Pro
$20 per month
Monthly
Extended Agent limits, frontier models, MCPs, skills and hooks, and cloud agents. Includes $20 of API agent usage plus bonus usage.
Pro+
$60 per month
Monthly
Three times the usage of Pro on OpenAI, Claude and Gemini models. Listed as Pro Plus in the pricing documentation, with $70 of included API agent usage.
Ultra
$200 per month
Monthly
Twenty times the usage of Pro on OpenAI, Claude and Gemini models, and priority access to new features. Includes $400 of API agent usage.
Teams
$40 per user per month
Monthly
Shared chats, commands and rules, centralised billing, org-wide Privacy Mode controls, SAML or OIDC SSO.
Enterprise
Custom
Custom
For invoicing, pooled usage and advanced security.
Figures are from Cursor's pricing page and pricing documentation. Cursor adjusts its plans often, and at least one indexed copy of the pricing page showed lower individual and team prices, so confirm current figures at cursor.com/pricing.
Prices change. Confirm current plans and terms on the vendor's site before you decide.
Visit official website (external) ↗See alternativesCursor integrations, API and platforms.
Important integrations
- MCP servers
- Skills and hooks
- OpenAI, Anthropic and Google models
- Cloud agents
- Bugbot
Supported platforms
- Desktop app (download from cursor.com)
API
Cursor says some features, including Tab and Apply from Chat, are powered by custom models and cannot be charged to a user's own API key.
Data import and export
Not reviewed for this profile.
Cursor says generated code belongs to the user regardless of plan and may be used commercially, per its pricing FAQ.
How Cursor works as an AI product.
Attributes are listed only where documented. We do not infer autonomy levels, certifications or model capabilities.
Classification: Agentic tool. A human-directed tool that can plan, edit and run actions, with the user reviewing or approving changes. Developer: Anysphere.
| Attribute | Details |
|---|---|
| Supported models | Models from OpenAI, Anthropic (Claude) and Google (Gemini), plus Cursor's own models. Cursor says it is powered by a combination of custom models and API models. [2] [1] |
| Tools and integrations | MCP servers, skills and hooks on Pro and above; Bugbot for code review; cloud agents. [1] |
| Ability to execute actions | Cursor offers an Agent and cloud agents. Third-party reviews describe agent mode editing files across a repository and running tests; vendor documentation on specific action types was not reviewed. [1] [4] |
| Execution environment | A desktop editor for local use, plus cloud agents that run remotely and are charged at model API pricing. [3] [1] |
| Human-in-the-loop controls | Not documented in the sources reviewed. |
| Permissions and access | Not documented in the sources reviewed. |
| Memory and context | Max Mode on certain models allows longer reasoning and context windows up to 1M tokens. Teams adds shared rules, commands and chats. [3] [1] |
| Deployment options | Desktop app. Subscriptions are sold directly through cursor.com. Enterprise is custom for invoicing, pooled usage and advanced security. [1] |
| Data handling and privacy | With Privacy Mode enabled, Cursor says code is never trained on and is retained only when required to operate a remote feature such as a background agent. Team admins can enforce Privacy Mode org-wide. [2] [1] |
| Security and compliance | SAML or OIDC SSO, an admin dashboard with usage stats and Privacy Mode enforcement are listed for team plans. Compliance certifications were not reviewed for this profile; check Cursor's security documentation. [3] |
| Pricing and usage limits | Hobby has limited Agent requests and Tab completions. Paid plans include usage charged at model API prices, and on-demand usage continues after the included amount, billed in arrears. [3] [1] |
| Languages and platforms | Not documented in the sources reviewed. |
| Limitations and known constraints | Tab and Apply from Chat use custom models and cannot be charged to a user's own API key. Included usage is finite, so heavy agent use can lead to on-demand charges. [2] [3] |
Primary tasks
- Writing and editing code across files
- Autocomplete
- Code review
- Running agents in the cloud
Best-fit workflows
- Developers who want an AI-native editor with the agent, completions and review in one place
- Teams that want several model vendors and org-wide privacy controls
- Workflows that benefit from cloud agents and agentic code review
Balanced and sourced.
Each point cites a source. Editorial summaries are not customer reviews.
Strengths
- Offers models from several vendors in one product, so teams are not tied to a single model family. [2]
- A free Hobby plan lets developers try the editor and agent before paying. [1]
- Team plans document org-wide Privacy Mode enforcement, SAML or OIDC SSO and an admin dashboard. [3]
- Certain models can use Max Mode with context windows up to 1M tokens, which is relevant to large codebases. [3]
Limitations
- Usage is metered at model API prices against an included amount, so heavy agent users may incur on-demand charges. [3]
- Tab and Apply from Chat run on custom models and cannot be billed to a user's own API key. [2]
- The plan and price structure changes often, and versions of the pricing page have differed, so budgeting requires checking the live page. [1]
- Human-approval and permission controls for the agent were not documented in the sources reviewed for this profile. [1]
Best for
- Developers who prefer an editor-centred workflow
- Teams that want to choose between several model vendors
- Organisations that want org-wide Privacy Mode and SSO
May be less suited to
- Developers who prefer to stay in a terminal-only workflow
- Teams that need documented agent permission controls before evaluating
Cursor compared with similar ai coding agents.
Summaries come from each profile. This is a side-by-side aid, not a ranking.
| Product | Interface | Model choice | Free tier | Pricing model | Human oversight |
|---|---|---|---|---|---|
| Cursor | AI code editor (desktop) with cloud agents | OpenAI, Anthropic, Google and Cursor models | Yes (Hobby) | Monthly tiers; usage at model API rates | Not documented in sources reviewed |
| Claude Code Compare | Terminal, VS Code, JetBrains, desktop app, web | Claude models (Anthropic) | Not documented in sources reviewed | Included in paid Claude plans, or usage-based API billing | Permissions, sandboxing and checkpoints documented |
Cursor questions and answers
What is Cursor used for?
Cursor is used to write, edit and review code with AI help inside an editor. It includes an Agent, Tab autocomplete, cloud agents and Bugbot code review.
Is Cursor an autonomous agent?
We classify it as an agentic tool. It offers an Agent and cloud agents, but vendor documentation reviewed here does not describe it as fully autonomous, and developers direct and review its work.
Does Cursor have a free plan?
Yes. The Hobby plan is free, needs no credit card, and includes limited Agent requests and limited Tab completions.
Which models does Cursor support?
Cursor lists models from OpenAI, Anthropic and Google, and uses its own custom models for features such as Tab. Model access and usage depend on the plan.
How does Cursor compare with Claude Code?
Cursor is an editor-centred product with multi-vendor models, while Claude Code is a terminal-first agentic tool from Anthropic that also runs in IDEs, a desktop app and the web. See the Cursor vs Claude Code comparison.
What integrations does Cursor support?
Documented integrations and ecosystem items include MCP servers, Skills and hooks, OpenAI, Anthropic and Google models, Cloud agents, Bugbot. API details: Cursor says some features, including Tab and Apply from Chat, are powered by custom models and cannot be charged to a user's own API key.
What are the limitations of Cursor?
Usage is metered at model API prices against an included amount, so heavy agent users may incur on-demand charges. Tab and Apply from Chat run on custom models and cannot be billed to a user's own API key. The plan and price structure changes often, and versions of the pricing page have differed, so budgeting requires checking the live page.
Which platforms does Cursor run on?
Cursor is documented as available on: Desktop app (download from cursor.com).
Is Cursor an AI agent or an AI assistant?
We classify Cursor as: Agentic tool. A human-directed tool that can plan, edit and run actions, with the user reviewing or approving changes. The classification is based on vendor documentation and is not an endorsement.
When was the Cursor profile last verified?
Sources for this profile were last reviewed on 15 Oct 2026. 4 sources are listed under Sources, and prices and plans can change, so confirm with the vendor.
Where this information comes from.
- Cursor pricing Cursor (cursor.com) Official Accessed 15 Oct 2026
- Cursor pricing FAQ Cursor (cursor.com) Official Accessed 15 Oct 2026
- Cursor documentation: Pricing Cursor (cursor.com/docs) Official docs Accessed 15 Oct 2026
- Cursor AI review Aixploria Third-party Accessed 15 Oct 2026 Used only for a third-party description of agent mode behaviour.
Further official pages: Cursor pricing (external)
Plans, usage rules and administration features were checked against Cursor's published pricing pages and documentation. Agent behaviour was not tested, and security certifications were not reviewed.
Stack18 has not tested this product or received anything from the vendor for this profile. Editorial summaries of vendor information are not endorsements. Confirm details with the vendor before buying.
Evaluating Cursor?
Check the vendor's current pricing and terms, compare it with similar products, or read how others approach the decision.
Logos provided by Logo.dev. Product names and logos are trademarks of their respective owners and appear for identification only; they do not imply endorsement.