ReasoningParams

class nemo_microservices.types.shared.ReasoningParams(*args: Any, **kwargs: Any)

Bases: BaseModel

effort: str | None = None

Option for OpenAI models to specify low, medium, or high reasoning effort which balances between speed and reasoning accuracy.

end_token: str | None = None

Configure the end token to trim reasoning context based on the model’s reasoning API. Used for omitting Nemotron reasoning steps from output denoted with </think> tags