Struct CascadeConfig 

pub struct CascadeConfig {
    pub quality_threshold: f64,
    pub max_escalations: u8,
    pub classifier_mode: CascadeClassifierMode,
    pub window_size: usize,
    pub max_cascade_tokens: Option<u32>,
    pub cost_tiers: Option<Vec<String>>,
}

Configuration for cascade routing (strategy = "cascade").

Cascade routing tries providers in chain order (cheapest first), escalating to the next provider when the response is classified as degenerate (empty, repetitive, incoherent). Chain order determines cost order: first provider = cheapest.

Limitations

The heuristic classifier detects degenerate outputs only, not semantic failures. Use classifier_mode = "judge" for semantic quality gating (adds LLM call cost).

Fields

quality_threshold: f64

Minimum quality score [0.0, 1.0] to accept a response without escalating. Responses scoring below this threshold trigger escalation.

max_escalations: u8

Maximum number of quality-based escalations per request. Network/API errors do not count against this budget. Default: 2 (allows up to 3 providers: cheap → mid → expensive).

classifier_mode: CascadeClassifierMode

Quality classifier mode: "heuristic" (default) or "judge". Heuristic is zero-cost but detects only degenerate outputs. Judge requires a configured summary_model and adds one LLM call per evaluation.

window_size: usize

Rolling quality history window size per provider. Default: 50.

max_cascade_tokens: Option<u32>

Maximum cumulative input+output tokens across all escalation levels. When exceeded, returns the best-seen response instead of escalating further. None disables the budget (unbounded escalation cost).

cost_tiers: Option<Vec<String>>

Explicit cost ordering of provider names (cheapest first). When set, cascade routing sorts providers by their position in this list before trying them. Providers not in the list are appended after listed ones in their original chain order. When unset, chain order is used (default behavior).

Trait Implementations

impl Clone for CascadeConfig

fn clone(&self) -> CascadeConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CascadeConfig

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

Formats the value using the given formatter.

impl Default for CascadeConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for CascadeConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for CascadeConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.