Reasoning model — slower but stronger at multi-step problems.
Reference this model in your agent's Agentfile:
Agentfile