Cursor gives you access to multiple frontier models — choosing the right one affects quality, speed, and cost.

Available Models

• claude-sonnet-4-6: Best overall for code — excellent reasoning, large context, Cursor's top recommendation for most tasks
• GPT-4o: Fast and capable — great for quick edits and chat
• o1 / o3-mini: Best for hard algorithmic problems and complex debugging
• cursor-small: Fast, free completions — used for Tab suggestions

Recommended Setup

• Default: claude-sonnet-4-6 for Chat and Composer
• Switch to o1 for complex debugging sessions
• cursor-small handles Tab automatically

Reference:

Cursor model selection

https://docs.cursor.com/settings/models