Cursor owns the IDE, Claude Code owns the terminal, Windsurf owns regulated government deals.
These three tools look identical on the surface but sell to completely different buyers. Cursor is the $29B platform play for Fortune 500 engineering orgs. Claude Code is the terminal-native agent for developers who already live in Anthropic's ecosystem. Windsurf is the only one with FedRAMP High authorization, making it the default choice for government and defense contracts.
Key takeaways
Winners by dimension
Side-by-side
| Cursor | Claude Code | Windsurf | |
|---|---|---|---|
| Free tier | Limited agent requests, limited Tab completions, no frontier models, no MCPs | Chat on web/iOS/Android, code generation, web search, no Claude Code included | Unlimited Tab autocomplete, SWE-1-mini model, light Cascade quota, MCP support |
| Starting paid price | $20/month (Pro) | $17/month annual or $20/month monthly (Pro, includes Claude Code) | $20/month (Pro) |
| Pricing model | Per-seat subscription with monthly usage credits; on-demand overage billed in arrears | Per-seat subscription with usage capacity tiers; API usage billed per token for Enterprise | Daily and weekly refreshing usage quota per tier; Tab autocomplete unlimited on all plans |
| Power user / max individual plan | Ultra at $200/month (20x usage multiplier) | Max 20x at $200/month (20x more usage than Pro) | Max at $200/month (heavy Cascade quota) |
| Teams plan | $40/user/month with SSO, RBAC, usage analytics, and centralized billing | $20/seat/month standard or $100/seat/month premium (5x usage); SSO included | $40/user/month with admin dashboard, priority support, pooled credits; no SSO |
| Enterprise compliance certifications | SOC 2 Type II, GDPR, CCPA | HIPAA-ready, SOC 2, SCIM, audit logs, compliance API | SOC 2 Type 2, FedRAMP High, HIPAA, ITAR |
| Proprietary AI model | Composer 2 (agentic coding) and Tab model (RL-trained autocomplete) | Claude Opus 4.7, Sonnet 4.6, Haiku 4.5 (all first-party Anthropic models) | SWE-1.5 (13x faster than Sonnet 4.5), SWE-1, SWE-1-mini |
| Context window | Not publicly specified; uses frontier model limits per model selected | 1M tokens on Opus 4.6 for Max, Team, and Enterprise plans | Not publicly specified; Fast Context auto-indexes codebases up to 1M+ lines |
| IDE flexibility | VS Code fork (desktop); JetBrains via ACP partnership | Terminal CLI; VS Code extension; JetBrains plugin; GitHub Actions | Standalone VS Code fork plus plugins for 40+ IDEs including all JetBrains, Vim, Neovim, Xcode |
| Code review / CI integration | Bugbot AI code review on GitHub PRs; GitHub Actions; Slack integration | PR generation, issue triage, GitHub Actions; no dedicated code review product | No dedicated code review product; Devin Cloud handles async tasks |
| Non-developer / knowledge worker features | None. Developer-only platform. | Claude Cowork (computer use), Excel/Word/PowerPoint add-ins, Chrome browser automation | None. Developer-only platform. |
| Open roles (growth signal) | 68 open roles across sales, engineering, ML research, and field marketing | 214 open roles; largest department is AI Research and Engineering with 71 roles | 15 open roles; focused on product growth, enterprise sales, and AI/ML engineering |
Who should pick whom
What we found
Three different buyers, not three competing products
Cursor sells to engineering organizations that want one platform for the entire software development lifecycle. Claude Code sells to developers and knowledge workers already inside Anthropic's ecosystem who want the best model without switching tools. Windsurf sells to regulated industries where FedRAMP High and ITAR certifications are table stakes. The overlap in the middle is real but smaller than it looks.
Pricing models reveal strategic intent
All three land at $20/month for a paid individual plan, but the mechanics differ sharply. Cursor uses usage credit multipliers (1x, 3x, 20x) that created a billing backlash in June 2025. Windsurf uses daily and weekly refreshing quotas, which feels more predictable. Claude Code ties usage to seat tier and keeps overage tied to API token pricing. Predictability wins developer trust. Cursor learned this the hard way.
Where the real moat is in 2026
Cursor's moat is network effects, not features. The SpaceX model training partnership, Bugbot's learned rules, and the Marketplace ecosystem create proprietary data loops that get harder to replicate over time. Windsurf's moat is compliance certifications. FedRAMP High takes 12 to 18 months to obtain. That's a real barrier. Claude Code's moat is Anthropic's model quality and the $380B valuation funding the compute to stay ahead.
What the hiring data tells you
Anthropic has 214 open roles and Cursor has 68. Windsurf has 15. That gap tells you everything about organizational stage. Anthropic is building a research and infrastructure company that happens to have a coding product. Cursor is scaling a go-to-market machine across verticals. Windsurf is a focused team executing on enterprise deals. Smaller headcount isn't weakness here. It's a signal of different ambition and burn rate.
Want to share this?
On paper they're all AI code editors at $20/month. In practice they sell to three completely different humans. Here's what the data actually shows:
Want this kind of report on your competitors?
ClientCues runs deep AI scans + side-by-side comparisons every week.
Try ClientCues Free