Windsurf's Cascade agent supports a choice of models. Claude Sonnet 4.6 and GPT-5.4 are the best-balanced everyday picks; Claude Opus 4.8 is the quality option for hard tasks; and Gemini 3.5 Flash or Grok 4.3 keep costs down without giving up much capability.
The strongest everyday driver for Cascade's multi-step coding flows.
OpenAI's value flagship — a great default if you prefer the GPT family in Windsurf.
For the hardest Cascade tasks where output quality is worth the premium.
Fast and cheap with a 1M context — keeps high-volume Cascade runs affordable.
Frontier-class quality at a friendlier price — a smart value choice.
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.