Struct AiModel 

pub struct AiModel {
    pub provider: String,
    pub provider_model: String,
    pub failover: Vec<AiModelTarget>,
    pub canary: Vec<AiCanaryVersion>,
    pub routing_policy: Option<String>,
    pub ab_test: Option<String>,
    pub canary_analysis: Option<CanaryAnalysis>,
}

One model alias mapping (DW-075 ai.models{alias}).

Fields

provider: String

Name of the ai.providers[] entry that serves this alias.

provider_model: String

The provider’s own model identifier (e.g. gpt-4o-mini, claude-sonnet-4-5, gemini-2.5-flash), sent to the provider in place of the alias.

failover: Vec<AiModelTarget>

Ordered failover chain (DW-076): when the primary target answers 429 or a 5xx, or its transport errors, the gateway retries the next entry — a DIFFERENT provider/model pair, never a re-send to the provider that just failed (its upstream’s breaker owns same-provider retries). The client sees one response. Absent (the default): no failover, the primary’s answer is final. Cannot be combined with canary on the same alias (validation rejects the pairing): failover is an availability chain for a single logical model; a canary split is deliberately serving multiple versions at once — mixing the two would retry a canary request onto the stable version and silently undo the experiment.

canary: Vec<AiCanaryVersion>

Weighted canary split (DW-076): traffic for this alias distributes across the listed versions by a deterministic weighted hash of the request id (the same slot semantics as traffic splitting: ratios hold per request, and re-sending a request with the same id lands on the same version). Each version names exactly one provider/model pair and carries no failover of its own. Absent (the default): no split. Cannot be combined with failover on the same alias.

routing_policy: Option<String>

A routing policy name (DW-085): when set, this alias is served by the named entry in ai.routing_policies instead of a plain failover chain or canary split. The policy is evaluated per request (a FallbackChain policy may call an external classifier service to decide cheap-vs-expensive; a LatencyCost policy picks deterministically). Cannot be combined with failover or canary on the same alias (validation enforces mutual exclusivity): a policy alias composes over OTHER aliases’ routing plans, so it has no primary provider/model pair of its own.

ab_test: Option<String>

An A/B test name (DW-086): when set, this alias is served by the named entry in ai.experiments.ab_tests instead of a plain failover chain, canary split, or routing policy. The experiment is evaluated per request (a deterministic weighted pick by request id selects a variant). Cannot be combined with failover, canary, or routing_policy on the same alias (validation enforces mutual exclusivity): an experiment alias composes over OTHER aliases’ routing plans, so it has no primary provider/model pair of its own.

canary_analysis: Option<CanaryAnalysis>

Auto-canary analysis (DW-091): when set, a background controller compares the canary (second version) vs the baseline (first version) on error rate or latency and adjusts the canary split weights automatically. Only valid when the canary has exactly 2 versions (baseline + canary). Weight changes are transient (Generation swap) and revert on config reload.

Trait Implementations

impl Clone for AiModel

fn clone(&self) -> AiModel

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for AiModel

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for AiModel

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl JsonSchema for AiModel

fn schema_name() -> Cow<'static, str>

The name of the generated JSON Schema.

fn schema_id() -> Cow<'static, str>

Returns a string that uniquely identifies the schema produced by this type.

fn json_schema(generator: &mut SchemaGenerator) -> Schema

Generates a JSON Schema for this type.

fn inline_schema() -> bool

Whether JSON Schemas generated for this type should be included directly in parent schemas, rather than being re-used where possible using the $ref keyword.

impl PartialEq for AiModel

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for AiModel

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for AiModel