Enum ReasoningEffort 

#[non_exhaustive]
pub enum ReasoningEffort {
    None,
    Minimal,
    Low,
    Medium,
    High,
    XHigh,
    Other(String),
}

A neutral reasoning-effort level, mapped per-wire (ADR-0007).

Grok Build proves this shape: one neutral enum with per-backend mappings (to_messages_api() → Anthropic budget, to_responses_api() → OpenAI effort). Tiers fragment per vendor/model generation (OpenAI none…xhigh, codex minimal…xhigh, grok stores per-model allowed-effort lists), so the ladder covers the observed union and Other passes vendor-specific strings through verbatim (ADR-0007 amendment 2026-07-19). Wires surface the API’s own error on unsupported tiers — never silently clamp (silent clamping would corrupt eval comparisons).

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

None

Explicitly send the wire’s “none” (distinct from omitting the param — SamplingArgs.reasoning_effort: None omits it entirely).

Minimal

Minimal reasoning.

Low

Low reasoning effort.

Medium

Medium reasoning effort.

High

High reasoning effort.

XHigh

Extra-high reasoning effort.

Other(String)

A vendor/model-specific tier, passed through verbatim by wires that take effort strings; wires with fixed mappings soft-reject it.

Trait Implementations

impl Clone for ReasoningEffort

fn clone(&self) -> ReasoningEffort

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ReasoningEffort

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Eq for ReasoningEffort

impl PartialEq for ReasoningEffort

fn eq(&self, other: &ReasoningEffort) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for ReasoningEffort