AI Models Reference

Guide to AI models available in Cursor and when to use each.

---

Model Overview

Anthropic Claude

Model	Description
Claude Opus 4	Most capable, complex reasoning
Claude Sonnet 4	Balanced speed/quality (recommended)
Claude Sonnet 3.5	Fast, efficient for most tasks

OpenAI GPT

Model	Description
GPT-4o	Multimodal, fast responses
GPT-4o mini	Cost-effective, quick tasks
o1-preview	Deep reasoning, complex problems
o1-mini	Fast reasoning model

Cursor Models

Model	Description
cursor-small	Optimized for Tab autocomplete

---

Claude Models (Anthropic)

Claude Opus 4

Attribute	Value
Best For	Complex reasoning, large refactors, architecture decisions
Speed	Slower
Quality	Highest
Context Window	200K tokens
Use When	Quality matters more than speed

Ideal Use Cases:

• System architecture design
• Complex debugging
• Large-scale refactoring
• Security audits
• Code review

---

Claude Sonnet 4 (Recommended)

Attribute	Value
Best For	Daily coding, balanced tasks
Speed	Fast
Quality	Excellent
Context Window	200K tokens
Use When	Most everyday coding tasks

Ideal Use Cases:

• Feature implementation
• Bug fixes
• Code generation
• Documentation
• General Q&A

---

Claude Sonnet 3.5

Attribute	Value
Best For	Quick tasks, simple edits
Speed	Very fast
Quality	Good
Context Window	200K tokens
Use When	Speed is priority

Ideal Use Cases:

• Quick fixes
• Simple refactors
• Code formatting
• Small features
• Rapid iteration

---

GPT Models (OpenAI)

GPT-4o

Attribute	Value
Best For	Multimodal tasks, image analysis
Speed	Fast
Quality	Excellent
Context Window	128K tokens
Use When	Need image understanding

Ideal Use Cases:

• UI/UX from screenshots
• Diagram interpretation
• Mixed media tasks
• General coding

---

GPT-4o Mini

Attribute	Value
Best For	Simple tasks, cost optimization
Speed	Very fast
Quality	Good
Context Window	128K tokens
Use When	Simple, repetitive tasks

Ideal Use Cases:

• Boilerplate generation
• Simple edits
• Quick questions
• Formatting

---

o1-preview

Attribute	Value
Best For	Complex reasoning, math, algorithms
Speed	Slower (reasoning time)
Quality	Exceptional for logic
Context Window	128K tokens
Use When	Complex problems needing deep thought

Ideal Use Cases:

• Algorithm design
• Mathematical problems
• Complex logic
• Multi-step reasoning
• Optimization problems

---

o1-mini

Attribute	Value
Best For	Faster reasoning tasks
Speed	Moderate
Quality	Excellent for logic
Context Window	128K tokens
Use When	Need reasoning but faster

Ideal Use Cases:

• Algorithm optimization
• Code analysis
• Logic problems
• Debugging complex issues

---

Model Selection Guide

By Task Type

Task	Recommended Model
Daily coding	Claude Sonnet 4
Complex architecture	Claude Opus 4
Quick fixes	Claude Sonnet 3.5
Image/UI work	GPT-4o
Algorithm design	o1-preview
Simple boilerplate	GPT-4o mini
Tab autocomplete	cursor-small (auto)

---

By Priority

Priority	Model Choice
Speed first	Claude Sonnet 3.5 or GPT-4o mini
Quality first	Claude Opus 4 or o1-preview
Balanced	Claude Sonnet 4 or GPT-4o
Cost first	GPT-4o mini or Claude Sonnet 3.5

---

Switching Models

In Chat

1. Quick Switch:

   • Press Cmd + / (macOS) or Ctrl + / (Windows)
   • Select model from dropdown

2. Per-Message:

   • Click model name at bottom of chat
   • Select for current message only

3. Default Model:

   • Settings → Models → Default Model
   • Changes default for new chats

---

In Inline Edit

• Uses same model selection as chat
• Can override per-edit with Cmd+/

---

For Tab Autocomplete

• Uses cursor-small automatically
• Optimized for speed and relevance
• Cannot be changed manually

---

Model Comparison Table

Model	Speed	Quality	Context	Cost
Claude Opus 4	⭐⭐	⭐⭐⭐⭐⭐	200K	$$$
Claude Sonnet 4	⭐⭐⭐⭐	⭐⭐⭐⭐	200K	$$
Claude Sonnet 3.5	⭐⭐⭐⭐⭐	⭐⭐⭐	200K	$
GPT-4o	⭐⭐⭐⭐	⭐⭐⭐⭐	128K	$$
GPT-4o mini	⭐⭐⭐⭐⭐	⭐⭐⭐	128K	$
o1-preview	⭐⭐	⭐⭐⭐⭐⭐	128K	$$$
o1-mini	⭐⭐⭐	⭐⭐⭐⭐	128K	$$

---

Request Limits

Pro Plan

Type	Limit
Fast requests	500/month
Slow requests	Unlimited
Premium requests (Opus/o1)	10/day

Business Plan

Type	Limit
Fast requests	1000/month
Slow requests	Unlimited
Premium requests	50/day

---

Using Your Own API Keys

Add Custom Keys

1. Settings → Models → API Keys
2. Enter your API key:
   • Anthropic API key for Claude
   • OpenAI API key for GPT

Benefits of Own Keys

• No request limits
• Pay-per-use pricing
• Access to all model versions
• Privacy control

Configuration

// Example: Add to Cursor settings
{
  "anthropic.apiKey": "sk-ant-...",
  "openai.apiKey": "sk-..."
}

---

Model Tips

Start Fast, Upgrade as Needed

• Begin with Sonnet for most tasks
• Switch to Opus when you need deeper analysis
• Use o1 for complex algorithmic problems

Match Model to Context Size

• Large codebases → Claude (200K context)
• Focused tasks → Any model works
• Many files referenced → Prefer Claude

Save Premium Requests

• Use Opus/o1 for important decisions
• Draft with Sonnet, refine with Opus
• Don't use premium for simple tasks

Experiment

• Try different models for same task
• Note which works best for your workflow
• Models evolve - retest periodically

---

Changelog

Models are updated regularly. Check Cursor release notes for:

• New model additions
• Performance improvements
• Context window changes
• Deprecated models

---

See Also

• Agent Chat SOP
• Agent Modes SOP
• VS Code Migration