Struct OpenRouterBackend 

pub struct OpenRouterBackend { /* private fields */ }

OpenRouter multi-provider gateway. Composes OpenAICompatibleBackend with the OpenRouter preset + slug-aware capability + tokens overrides.

Implementations

impl OpenRouterBackend

pub fn from_env() -> Self

Construct from env. OPENROUTER_API_KEY is read at construction time; None is permitted (auth check fires at first call).

pub fn with_api_key(api_key: Option<String>) -> Self

Construct with an explicit API key (or None).

pub fn with_base_url(self, base_url: impl Into<String>) -> Self

Override the base URL (test fixtures, regional OpenRouter endpoints).

pub fn with_default_model(self, model: impl Into<String>) -> Self

Override the default slug. Useful when an adopter pins a specific model (e.g. anthropic/claude-haiku-4-5 for cheap drafting).

pub fn inner(&self) -> &OpenAICompatibleBackend

Borrow the underlying OpenAICompatibleBackend (for testing fixtures that need access to the composed inner state).

Trait Implementations

impl Backend for OpenRouterBackend

fn name(&self) -> &str

Short provider name used as the registry key. E.g. "anthropic", "openai", "kimi".

fn default_model(&self) -> &str

Default model used when ChatRequest::model is empty.

fn complete<'life0, 'async_trait>( &'life0 self, request: ChatRequest, ) -> Pin<Box<dyn Future<Output = Result<ChatResponse, BackendError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Synchronous-result chat completion (non-streaming).

fn stream<'life0, 'async_trait>( &'life0 self, request: ChatRequest, ) -> Pin<Box<dyn Future<Output = Result<ChatStream, BackendError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Streaming chat completion. Adopter consumes the returned stream; per-chunk text arrives in ChatChunk::delta, finish reason + usage in the final chunk.

fn count_tokens(&self, model: &str, text: &str) -> usize

Best-effort token count for text against a specific model on this provider. Default impl delegates to the unified tokens::count_tokens dispatch; per-provider overrides may consult the provider’s HTTP count_tokens endpoint when an exact answer is required + a network round-trip is acceptable.

fn supports(&self, capability: Capability, model: &str) -> bool

Capability discovery — does this backend support capability for the given model? Default returns false for everything; per-provider impls override.

impl Default for OpenRouterBackend

fn default() -> Self

Returns the “default value” for a type.