Description
Cursor
Overview
What It Does
- Tab: Predictive multi-line autocomplete
- Agent: Executes complex tasks autonomously
- Chat: Conversation with project context
- Cmd+K: Inline edits with natural language
Available Models
| Provider | Models |
|---|---|
| OpenAI | GPT-5, GPT-4.1 |
| Anthropic | Claude Sonnet 4.5, Claude Opus 4.5 |
| Gemini 3 Pro, Gemini Flash 2.5 | |
| xAI | Grok Code |
| Cursor | Composer (proprietary) |
Composer Model (Oct 2025)
- 4x faster than similar models
- Completes tasks in <30 seconds
- Trained with semantic search on codebases
- Mixture-of-Experts architecture
- ~250 tokens/second
Main Features
| Feature | Description |
|---|---|
| Tab Completion | Predicts next action, 28% higher acceptance |
| Agent Mode | Executes multi-step tasks, terminal, multi-file edits |
| Background Agents | Agents running in background |
| Multi-Agent | Up to 8 agents in parallel (git worktrees) |
| Codebase Indexing | Understands your entire project |
| Plan Mode | Plans before executing |
| Debug Mode | Reproduces and fixes bugs |
| Browser Tool | Integrated browser testing |
| Voice Control | Voice control |
| Privacy Mode | Zero-retention, SOC 2 |
Pricing (2025)
Individual Plans
| Plan | Price | Includes |
|---|---|---|
| Hobby | $0 | 1-week trial, limited Agent/Tab |
| Pro | $20/mo | Unlimited Tab, Extended Agent, Background Agents |
| Pro+ | $60/mo | 3x usage on OpenAI/Claude/Gemini |
| Ultra | $200/mo | 20x usage, priority features |
Business Plans
| Plan | Price | Includes |
|---|---|---|
| Teams | $40/user/mo | SSO, analytics, shared rules, privacy mode |
| Enterprise | Custom | SCIM, audit logs, pooled usage, invoice billing |
Bugbot Add-on
| Plan | Price |
|---|---|
| Free | $0 (limited reviews) |
| Pro | $40/user/mo (200 PRs/mo) |
| Teams | $40/user/mo (unlimited) |
Usage-Based System (Jun 2025)
- Pro includes $20 credit pool/month for premium models
- Tab and Auto mode: unlimited on Pro
- When credits run out: slow requests or pay-per-use
- "Thinking" models (Claude Sonnet) = 2 requests
Integrations
- VS Code: 100% compatible (extensions, themes, keybindings)
- GitHub: Bugbot integration
- Slack: Start tasks from Slack
- CLI: Cursor from terminal
- MCP: Connect external tools
Platforms
- macOS
- Windows
- Linux
PROS ✅
- VS Code base - Frictionless migration
- Multi-model choice - OpenAI, Anthropic, Google, xAI
- Composer model - Ultra-fast, designed for coding
- Codebase understanding - Understands huge projects
- Agent mode - Real autonomous tasks
- Tab accuracy - 28% higher acceptance than before
- Privacy Mode - SOC 2, zero-retention
- Fortune 500 trust - Used by top companies
- Active development - Weekly updates
CONS ❌
- Confusing pricing - Credit pools, usage-based
- Resource intensive - Can be heavy
- Credit anxiety - Runs out fast with premium models
- Pricing changes - Model changed several times
- AI hallucinations - Incident with "Sam" bot inventing policies
- Learning curve - Many new features
- Verification needed - Always review generated code
Why Choose Cursor?
- You're a professional developer
- You work on large codebases
- You want real agent mode
- You already use VS Code
- You need multiple AI models
- Your company requires SOC 2/Privacy
- Very limited budget
- You only need basic autocomplete
- You don't want to depend on AI
- You prefer simple predictable pricing
vs Competitors
| vs | Cursor Wins | Competitor Wins |
|---|---|---|
| GitHub Copilot | Multi-model, Agent, codebase context | Cheaper ($10), GitHub integration |
| Windsurf | More mature, Fortune 500 | Simpler, cascade mode |
| Claude Code | Complete IDE | Terminal-first, Anthropic native |
| Tabnine | More features | On-premise, offline |
Notable Customers
Metrics (2025)
- $1B+ ARR (Nov 2025)
- $29.3B valuation (Nov 2025)
- 1M+ DAU
- 50K+ businesses
- 300+ employees
- 9,900% YoY growth
Key Features
Built on VS Code with full compatibility
Predictive multi-line Tab Completion
Agent Mode for autonomous tasks
Proprietary Composer model 4x faster
Multi-Agent up to 8 agents in parallel
Background Agents running in background
Complete Codebase Indexing
Plan Mode to plan before executing
Debug Mode to reproduce and fix bugs
Browser Tool for integrated testing
Voice Control
Privacy Mode SOC 2 zero-retention
Multiple models OpenAI Anthropic Google xAI
Chat with full project context
Cmd+K inline edits with natural language
Bugbot for automatic code reviews
CLI for terminal use
Slack and GitHub integration
Git worktrees for agent isolation
Custom rules and commands per team
Use Cases
Professional software development
Large codebase refactoring
Context-aware code generation
AI-assisted debugging
Automatic code reviews
Complete feature development
Automated testing
Legacy code migration
Code documentation
Enterprise team development
Rapid prototyping
Complex project maintenance
Full-stack development
API creation
Frontend development React Vue Angular
Backend development Node Python Go
DevOps and scripting
Startups accelerating development
Agencies delivering projects
Freelancers increasing productivity
Information
Company
Anysphere
Website
cursor.com