Declares how a model exposes reasoning effort. Used as the single source of truth for resolving a
requested AiReasoningEffort onto what the model natively accepts. When a request cannot be
matched or clamped, no reasoning parameter is emitted and the model applies its own native default.
none: the model has no reasoning knob (e.g. it always reasons).
levels: the model accepts a fixed set of named levels (emitted as a string).
budget: the model accepts a token budget; each canonical level maps to a concrete integer.
Declares how a model exposes reasoning effort. Used as the single source of truth for resolving a requested AiReasoningEffort onto what the model natively accepts. When a request cannot be matched or clamped, no reasoning parameter is emitted and the model applies its own native default.
none: the model has no reasoning knob (e.g. it always reasons).levels: the model accepts a fixed set of named levels (emitted as a string).budget: the model accepts a token budget; each canonical level maps to a concrete integer.