Cursor lets you choose the model behind its agent and chat. Claude Sonnet 4.6 is the most popular daily driver; Claude Opus 4.8 and GPT-5.5 are the picks for the hardest tasks; Gemini 3.1 Pro shines on huge-context work; and Grok 4.3 offers frontier-class quality at better value.
The default most Cursor users keep selected — excellent agentic editing at a sensible price and speed.
For the hardest changes where output quality matters most — Cursor's heavy-lifting option.
OpenAI's flagship for top-tier reasoning, when you prefer the GPT family in Cursor.
The pick for very large contexts — holds big codebases for whole-repo reasoning.
Frontier-class quality at a friendlier price — a strong value choice in Cursor.
The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.