Models
Models
Pick the model per task and switch mid-task without losing history.
#Switch models
#List from the CLI
bash
codey modelsList every available model before pinning one in config.
#Switch mid-task
Open the /models menu in the terminal UI to swap to a cheaper or stronger model mid-task.
#Override per project
json
{
"model": "anthropic/claude-sonnet-4-5",
"small_model": "anthropic/claude-haiku-4-5",
"provider": {
"anthropic": {
"options": {
"timeout": 60000
}
}
}
}Pin the primary model, reserve the small model for fast low-cost tasks like title generation, and set the provider timeout in the same block.
#Tune reasoning
json
{
"agent": {
"deep-thinker": {
"model": "openai/o1",
"reasoningEffort": "high",
"maxSteps": 10
}
}
}Give thinking models a reasoning effort and a step cap per agent. Raise effort for harder tasks; lower the cap to bound cost.