52 facts per product, each taken from the vendor's own public pages and linked to its source. 13 rows differ and are highlighted; the better documented value in a row is bold. “Not documented” means the vendor's pages do not state it — it is never counted as a “No”.
Facts verified · Prices in USD · Machine-readable JSON
Cursor's Hobby plan is free with no credit card, and covers limited Agent requests and a monthly Tab allowance. Claude's Free plan does not include Claude Code. Paid entry is close: Cursor Pro is $16 per month billed yearly or $20 monthly; Claude Pro is $17 or $20.
Cursor is a desktop editor that runs VS Code extensions, with Tab autocomplete unlimited on paid plans and a local index for code search. Claude Code documents no as-you-type completion: it works through prompts and agent edits, and it searches files at run time instead of keeping an index.
Claude Code has a VS Code extension and a JetBrains plugin. Cursor's pages document no VS Code extension: Cursor is its own editor, which runs VS Code extensions. In JetBrains IDEs its agent runs through the Agent Client Protocol in the AI Chat panel, needs a paid Cursor plan, and does not list Tab autocomplete.
Claude Code's Team plan is $20 per seat per month billed annually or $25 monthly, for 2 to 150 seats. Cursor's Teams plan is $32 per user per month billed yearly or $40 monthly. Both include SSO and an admin dashboard with usage analytics.
Cursor offers its own Grok and Composer models alongside OpenAI GPT, Anthropic Claude, Google Gemini and others, or Auto routing. Claude Code runs Anthropic's Claude models only and does not support routing to other models.
Cursor's Privacy Mode is available on every plan, including free; with it on, Cursor's zero data retention agreements with model providers apply and Cursor does not train on your code. Claude Code offers zero data retention only to qualified Enterprise accounts, and Free, Pro and Max accounts are used for training while the user's model-improvement setting is on.
Claude Code's self-hosted environments, in public beta on Team and Enterprise, run whole cloud sessions on your infrastructure, with model inference still going to Anthropic, and the CLI runs locally. Cursor states it does not offer on-premises deployment; its Self-Hosted Machines run only the cloud agents' tool execution, while Cursor runs the agent loop and inference.
Both document SOC 2 Type II and ISO 27001:2022, and both sign HIPAA BAAs on Enterprise only: Cursor with Privacy Mode locked on org-wide, Claude Code with zero data retention on a qualified account. Both offer SSO from their team plan.
| Cursor | Claude Code | |
|---|---|---|
| Free plan | Yes Hobby plan; no credit card required. | No Claude's Free plan does not include Claude Code; Claude Code starts on Pro. The Console route is pay-as-you-go API usage. |
| Free plan limits | Limited Agent requests; access to Composer; monthly Tab allowance Help center: Hobby covers Agent, Chat and Tab completions with the Auto model and a smaller model set; Tab suggestions pause once the monthly allowance is used; free accounts cannot start Cloud Agents. | No free plan Claude's Free plan does not include Claude Code. |
| Free trial (days) | Not documented Not documented on vendor pages | Not documented Not documented on vendor pages. The plan comparison lists trials only for sales-assisted Enterprise, with no length stated. |
| Trial requires credit card | Not documented Not documented on vendor pages | Not documented Not documented on vendor pages |
| Cheapest paid plan, billed annually (per user / month) | $16.00 Pro (individual plan), billed yearly. An India-only Start plan (₹649/mo, billed monthly in INR) is listed in the help center, not on the pricing page. | $17.00 Pro plan, billed annually ($200 up front). |
| Cheapest paid plan, billed monthly (per user / month) | $20.00 Pro (individual plan), billed monthly. An India-only Start plan (₹649/mo, INR) is listed in the help center, not on the pricing page. | $20.00 Pro plan, billed monthly. |
| Highest listed-price plan, billed annually (per user / month) | $160.00 Ultra (individual plan), billed yearly. Highest per-seat team price is Teams Premium at $96/user/mo billed yearly. | $100.00 Team Premium seat. Max 20x ($200/month) costs more but is billed monthly only; Enterprise is $20/seat plus usage at API rates. |
| Minimum paid seats | Not documented Not documented on vendor pages | 2 Team plan is for 2 to 150 seats. Pro and Max are single-user plans. |
| Custom-priced enterprise plan | Yes Enterprise is contact-sales; priced per seat with an included usage allotment and optional pre-committed usage. | Yes Enterprise lists US$20/seat/month billed annually plus usage at API rates; the sales-assisted route gives a tailored quote. |
| Cursor | Claude Code | |
|---|---|---|
| Web app | Yes Browser access at cursor.com/agents for starting, following and reviewing cloud agents. | Yes Claude Code on the web (claude.ai/code) runs cloud sessions; Pro, Max, Team, and Enterprise premium or Chat + Claude Code seats. |
| Runs on Windows | Yes Desktop app for Windows x64 and ARM64 (user or system installer); CLI via WSL or native PowerShell. | Yes CLI on Windows 10 1809+ or Windows Server 2019+ (native or WSL); Desktop app also on Windows. |
| Runs on macOS | Yes Desktop app for Apple silicon, Intel and Universal builds; CLI via install script. | Yes macOS 13.0 or later. |
| Runs on Linux | Yes Desktop app as .deb, RPM and AppImage for x64 and ARM64; CLI via install script. | Yes CLI on Ubuntu 20.04+, Debian 10+, Alpine Linux 3.19+. The Desktop app on Linux is in beta. |
| iOS app | Yes Cursor for iOS on iPhone and iPad (iOS/iPadOS 26.0+) directs and reviews cloud agents and Remote Control sessions; no editor or terminal. Starting agents needs a paid plan. | Yes The Claude app is a client for cloud sessions and Remote Control of local sessions; code does not run on the phone. |
| Android app | No No Cursor app for Android; an Android version is planned with no release date. The separate Grok Bot app (a general-purpose agent included with paid Cursor plans) is on Google Play. | Yes The Claude app is a client for cloud sessions and Remote Control of local sessions; code does not run on the phone. |
| Browser extension | Not documented Not documented on vendor pages | Yes Claude in Chrome extension (Chrome, Edge, other Chromium browsers) lets Claude Code drive the browser for testing; requires a claude.ai login, not an API key. |
| Cursor | Claude Code | |
|---|---|---|
| Own standalone code editor | Yes Desktop app with an agent-first Agents Window plus a classic editor that runs VS Code extensions from the Open VSX registry and imports VS Code settings, keybindings and extensions. The vendor pages do not call it a fork. | Yes Claude Desktop's Code tab: an agent app with a file-editor pane, diff viewer and terminal. Not a full IDE and not a fork of another editor. |
| Extension for VS Code | Not documented Not documented on vendor pages | Yes Requires VS Code 1.94.0 or later. |
| Plugin for JetBrains IDEs | Yes Through the Agent Client Protocol: install Cursor from the ACP registry in the JetBrains AI Chat panel (AI Assistant plugin, 2025.1+). Paid Cursor plan required. Provides the agent (model choice, file edits, terminal commands); Tab autocomplete is not listed. | Yes IntelliJ IDEA, PyCharm, Android Studio, WebStorm, PhpStorm, GoLand and others; the plugin runs the CLI in the IDE terminal. |
| Command-line (terminal) agent | Yes Cursor CLI (`agent`): interactive, Plan and Ask modes plus headless mode for scripts, CI and GitHub Actions. Installs on macOS, Linux, Windows (WSL and native PowerShell). | Yes The CLI is the most complete surface; scripting and the Agent SDK are CLI-only. |
| Other editors supported | JetBrains IDEs (via ACP); any editor or IDE through the Cursor CLI in a terminal; custom ACP clients The Xcode integration is an MCP bridge that lets Cursor's agent build, test and edit Xcode projects from Cursor; it is not a plugin inside Xcode. | Cursor, Devin Desktop, Kiro and other VS Code forks (via the VS Code extension); any editor's terminal via the CLI Cursor has its own install link on the same page. |
| Cursor | Claude Code | |
|---|---|---|
| Inline code completion (autocomplete) | Yes Tab. Unlimited on Pro, Pro+ and Ultra; Hobby has a monthly Tab allowance. Tab always uses Cursor's own models, even with your own API keys. | No Stated for the Desktop app; no Claude Code surface documents as-you-type completion. It works through prompts and agent edits. |
| Agent that edits across multiple files | Yes Agent, Plan, Ask and Debug modes in the editor, Agents Window, CLI, web and iOS. | Yes |
| Agent runs terminal commands | Yes Run Mode sets when commands run automatically, ask for approval, or run in a sandbox. | Yes Asks permission before running commands unless a permission mode allows it. |
| Cloud / background agents | Yes Cloud Agents run in isolated cloud VMs, in parallel, from desktop, web, iOS, Slack, GitHub and Linear; Automations run agents on schedules or triggers. Included on paid plans (Pro and up); free accounts cannot start them. Tool execution can run on Self-Hosted Machines. | Yes Cloud sessions on Pro, Max and Team, and for Enterprise users with premium or Chat + Claude Code seats; not available through third-party providers. |
| Pull request code review | Yes Bugbot reviews PRs automatically or on demand. Usage-based billing: individual plans draw from included usage, then on-demand; Teams bill from on-demand spend. Not included in the India-only Start plan. | Yes Managed GitHub PR review is a research preview on Team and Enterprise, billed separately through usage credits (about $15-25 per review). Other plans can review a diff locally with /code-review. |
| Codebase indexing / semantic code search | Yes Local Instant Grep index for exact and regex code search, plus an Explore subagent for broad searches. Cursor states it does not upload code or store codebase embeddings for search. | No Uses agentic search (file-pattern and regex search) at run time instead of a code index. |
| MCP (Model Context Protocol) servers | Yes Pricing page lists MCPs, skills and hooks among Pro features; Enterprise admins can allowlist or block MCP servers. | Yes |
| Project rules / custom instruction files | Yes Project rules in .cursor/rules (.mdc files), AGENTS.md, user rules, and enforceable team rules on Teams and Enterprise. | Yes CLAUDE.md files and .claude/rules/; also reads AGENTS.md. |
| Checkpoints to roll back agent edits | Yes Created automatically before significant Agent changes; stored locally, separate from Git; restoring reverts files only. | Yes Rewind with /rewind or Esc twice. Files changed by Bash commands are not tracked. |
| Cursor | Claude Code | |
|---|---|---|
| User can choose the AI model | Yes Pick a model or use Auto (Cursor Router). Hobby gets a smaller model set; paid plans unlock all models; Enterprise admins can restrict models. | Yes Switch with /model or claude --model. |
| Model families offered | Cursor models (Grok 4.7, 4.6, 4.5; Composer 2.5), OpenAI GPT, Anthropic Claude, Google Gemini, and others Grok models are listed as Cursor models in the Cursor Models usage pool. The docs model table also lists Meta, Moonshot and Z.ai models. | Anthropic Claude only: Fable, Opus, Sonnet, Haiku Opus requires a paid plan; Fable draws on usage credits on Pro and 50% of weekly limits on Max. Non-Claude models are not supported. |
| Bring your own API key | Yes OpenAI, Anthropic, Google, Azure OpenAI and AWS Bedrock keys; chat models only (Tab stays on Cursor models). Requests still pass through Cursor's backend. Individual plans: not counted against included usage. Teams/Enterprise: $0.25 per million tokens Cursor Token Rate still applies. Enterprise admins can disable BYOK. Cursor's zero data retention does not cover BYOK requests. | Yes Anthropic Console API key, or your own Amazon Bedrock, Google Cloud Agent Platform or Microsoft Foundry account; usage billed by that provider. |
| Local / self-hosted models | Not documented Not documented on vendor pages | No Claude models only, served by the Anthropic API, Amazon Bedrock, Google Cloud Agent Platform or Microsoft Foundry. |
| How paid-plan usage is limited | Monthly included model usage in two pools (Cursor Models; Other Models at API prices), then optional on-demand pay-as-you-go at API rates Pro+ has 3x and Ultra 20x Pro Agent limits; Teams Premium has 5x Standard. Usage resets monthly without rollover. Teams and Enterprise add a $0.25 per million tokens Cursor Token Rate on third-party models. Enterprise offers pooled usage. | Usage pool shared with Claude chat: limits reset every 5 hours, plus weekly caps; optional usage credits at API rates Max = 5x or 20x Pro usage; Team Premium = 5x Standard. Enterprise and Console usage is billed per token at API rates. |
| Open-source client | No Proprietary client. The public github.com/cursor/cursor repository holds a README and issue templates, not the app source. Some auxiliary repos are open source (e.g. plugins, SDK bridge, Terraform provider). | No The public GitHub repo holds issues, changelog, plugins and examples; the client is proprietary. |
| Cursor | Claude Code | |
|---|---|---|
| Zero data retention option | Yes Applies with Privacy Mode on, available on every plan including free; on by default for Enterprise, enforceable org-wide by Teams and Enterprise admins. Exceptions: your own API keys, models that require provider retention (admin approval needed), and prompts flagged by providers' abuse classifiers. | Yes Enterprise only, for qualified accounts; not default and not in the standard Enterprise plan. Anthropic enables it per organization after an eligibility check. |
| Code excluded from model training | Yes Requires Privacy Mode. With Privacy Mode off, Cursor may use code data to train its models. Terms: Anysphere does not train on your content unless you explicitly agree. | Yes Team, Enterprise, API and cloud-platform use. Free, Pro and Max accounts are used for training when the user's model-improvement setting is on (opt-out). |
| Self-hosted or on-premises deployment | No No on-premises deployment; the service runs on AWS. Cloud Agent tool execution can move to Self-Hosted Machines you manage, while Cursor still runs the agent loop and inference. AWS PrivateLink and Cloudflare Tunnel private connectivity are available. | Yes Self-hosted environments run cloud sessions on your own infrastructure (public beta, Team and Enterprise); model inference still goes to api.anthropic.com. The CLI itself runs locally. |
| Team admin dashboard and usage analytics | Yes Teams plan and up: centralized billing and administration with usage analytics. Enterprise adds SCIM, audit logs, AI Code Tracking API and Conversation Insights. | Yes Team and Enterprise at claude.ai/analytics/claude-code (Admins and Owners), and Console for API customers; seat management on Team and Enterprise. |
| Cursor | Claude Code | |
|---|---|---|
| Listed integrations | 370 Plugins listed in the Cursor Marketplace (MCP servers, skills and tool bundles from third parties and Cursor), counted as unique plugin listings; excludes 27 automation templates. Separately, the docs cover 11 native integrations: GitHub, GitLab, Azure DevOps, Bitbucket, Slack, Microsoft Teams, Jira, Linear, Notion, JetBrains and Xcode. | 313 Plugins in the Claude Marketplace directory filtered to 'Works with: Claude Code' (bundles of tools, skills and MCP integrations); 337 across Claude and Claude Code. |
| Public API | Yes Cloud Agents API (beta, all plans), TypeScript and Python SDKs (all users); Admin, Analytics, AI Code Tracking and Bugbot APIs for Enterprise teams. | Yes Agent SDK (CLI, Python and TypeScript packages). |
| Single sign-on (SSO) | Yes SAML/OIDC SSO on Teams and Enterprise at no extra cost; SCIM provisioning on Enterprise. | Yes Team and Enterprise plans; not on Pro or Max. |
| Two-factor authentication | Not documented Not documented on vendor pages | Not documented Not documented on vendor pages |
| SOC 2 report | Yes SOC 2 Type II; report available on request via trust.cursor.com. | Yes SOC 2 Type I and Type II; report available from the Anthropic Trust Center. |
| ISO 27001 certification | Yes ISO/IEC 27001:2022; also ISO/IEC 42001:2023 and AIUC-1. | Yes ISO 27001:2022; also ISO/IEC 42001:2023. |
| HIPAA compliance offered | Yes Enterprise only: signed BAA required, with Privacy Mode enabled and locked org-wide. The standard Terms of Service prohibit sending HIPAA-regulated data otherwise. | Yes Enterprise only. Claude Code is under the BAA only when zero data retention is enabled on a qualified account. |
| Cursor | Claude Code | |
|---|---|---|
| Support channels | Community forum, email (hi@cursor.com), help center with a signed-in support request form; priority support on Enterprise Individual and Teams plans: community and standard support. Enterprise: first human response within 8 hours (critical) or 24 hours (standard). | In-app support messenger (Fin AI bot, then Product Support team) and email No phone or live-chat support. Sales-assisted Enterprise adds customer success at certain spend thresholds. |
| 24/7 support | Not documented Not documented on vendor pages | Not documented Not documented on vendor pages |
| Vendor (legal entity) | Anysphere, Inc. Cursor's blog announced on Aug 14, 2026 that Cursor has been acquired by SpaceX; the Terms of Service still name Anysphere, Inc. | Anthropic, PBC Contracting entity is Anthropic Ireland, Limited for customers in the EEA, Switzerland and the UK. |
| Founded | Not documented Not documented on vendor pages | 2021 |
| Headquarters | San Francisco, CA, USA Correspondence address in the Terms of Service: 2261 Market Street STE 86466, San Francisco, CA 94114. The vendor pages do not name a headquarters. | San Francisco, California, United States Registered address per the privacy policy, which also states the company is headquartered in the United States. |
| Cursor plan | Billed annually | Billed monthly |
|---|---|---|
| Hobby Per month Free individual plan. No credit card required; limited Agent requests; access to Composer. | Free | Free |
| Pro Per month Individual plan. $16/mo billed yearly, $20/mo billed monthly. Extended limits on Agent, access to frontier models, MCPs, skills and hooks, Cloud agents, Grok Bot access; Bugbot on usage-based billing. Includes monthly usage in two pools (Cursor Models; Other Models at API prices), then optional on-demand usage at API rates. | $16.00 | $20.00 |
| Pro+ Per month Individual plan. $48/mo billed yearly, $60/mo billed monthly. Everything in Pro, plus 3x Pro limits on Agent and higher Grok Bot usage limits. | $48.00 | $60.00 |
| Ultra Per month Individual plan. $160/mo billed yearly, $200/mo billed monthly. Everything in Pro, plus 20x Pro limits on Agent, highest Grok Bot usage limits and priority access to new features. | $160.00 | $200.00 |
| Teams (Standard seat) $32/user/mo billed yearly, $40/user/mo billed monthly. Centralized team billing and administration, team marketplace for rules, skills and plugins, cloud agents and automations with shared team context, agentic code reviews with Bugbot, usage analytics, team-wide privacy mode, SAML/OIDC SSO. Per-seat included usage; on-demand usage beyond it. | $32.00 | $40.00 |
| Teams (Premium seat) $96/user/mo billed yearly, $120/user/mo billed monthly. Same Teams features with 5x Standard limits on Agent and higher Grok Bot usage limits. Standard and Premium are seat types within the Teams plan. | $96.00 | $120.00 |
| Enterprise Custom (contact sales) Everything on Teams plus pooled usage, invoice/PO billing, SCIM seat management, repository/model/MCP access controls, auto-run, browser and network controls, audit logs and service accounts, AI code tracking API, priority support and account management. | Custom | Custom |
| Claude Code plan | Billed annually | Billed monthly |
|---|---|---|
| Pro Per month Individual plan; Claude Code included. Annual billing is $200 up front. Usage limits shared with Claude chat. | $17.00 | $20.00 |
| Max 5x Per month Individual plan; 5x the usage of Pro per 5-hour session. Billed monthly only. | Not offered | $100.00 |
| Max 20x Per month Individual plan; 20x the usage of Pro per 5-hour session. Billed monthly only. | Not offered | $200.00 |
| Team (Standard seat) Teams of 2 to 150; Claude Code included; more usage than Pro. SSO, central billing, usage analytics. | $20.00 | $25.00 |
| Team (Premium seat) 5x more usage than Standard seats; seat types can be mixed within one Team plan. | $100.00 | $125.00 |
| Enterprise Per user / month + usage at API rates Seat price plus all usage billed at API rates; billed annually only. Self-serve or sales-assisted (tailored quote). Adds SCIM, audit logs, custom data retention, role-based access. | $20.00 | Not offered |
| Claude API (Console) Pay as you go (per million tokens) Claude Code billed at standard API token prices through a Claude Console account; no per-seat fee. Listed API prices per million input/output tokens: Haiku 4.5 $1/$5, Sonnet 5 $2/$10, Opus 5.5 $4/$20, Fable 5.1 $10/$50. | Pay as you go | Pay as you go |
Per user per month unless the plan states otherwise, as listed on each vendor's pricing page.