DeepSeek Kotlin SDK/org.oremif.deepseek.models.responses/ReasoningConfig

ReasoningConfig

@Serializable

class ReasoningConfig(val effort: ReasoningEffort? = null)

Thinking-mode configuration of a Responses API request.

Unlike /chat/completions, which splits the toggle (org.oremif.deepseek.models.Thinking) from the budget (reasoning_effort), the Responses API folds both into effort: ReasoningEffort.NONE switches thinking off, and every other level switches it on with that budget. Left unset, the model's own default applies — thinking on.

Example:

val quick = responseParams { reasoning = ReasoningConfig(ReasoningEffort.NONE) }
val deep = responseParams { reasoning = ReasoningConfig(ReasoningEffort.MAX) }

The API accepts summary as well but never generates one, so the SDK leaves it out.

Constructors

ReasoningConfig

constructor(effort: ReasoningEffort? = null)

Properties

effort

val effort: ReasoningEffort?

How much reasoning the model spends before answering.

Functions

equals

open operator override fun equals(other: Any?): Boolean

hashCode

open override fun hashCode(): Int

toString

open override fun toString(): String