Struct Client 

pub struct Client { /* private fields */ }

Asynchronous API client for querying OpenAI and OpenRouter endpoints.

Implementations

impl Client

pub fn new(api_key: impl Into<String>) -> Result<Self, ClientError>

Create a new client targeting OpenAI with an API key.

pub fn openai(api_key: impl Into<String>) -> Result<Self, ClientError>

Create a client targeting the official OpenAI API.

pub fn openrouter(api_key: impl Into<String>) -> Result<Self, ClientError>

Create a client targeting the OpenRouter API.

pub fn openrouter_with_attribution( api_key: impl Into<String>, app_url: impl Into<String>, app_name: impl Into<String>, ) -> Result<Self, ClientError>

Create a client targeting OpenRouter with application attribution metadata.

pub fn custom( base_url: impl Into<String>, api_key: Option<String>, ) -> Result<Self, ClientError>

Create a client targeting a custom OpenAI-compatible server.

pub fn from_env() -> Result<Self, ClientError>

Automatically configure a client from available environment variables.

Checks OPENROUTER_API_KEY first; if present, connects to OpenRouter. Then checks OPENAI_BASE_URL; if present, connects to that custom endpoint with optional OPENAI_API_KEY. Otherwise checks OPENAI_API_KEY to connect to OpenAI.

pub fn builder(provider: Provider) -> ClientBuilder

Initialize a builder with a specific provider.

pub fn provider(&self) -> &Provider

Returns a reference to the active provider.

pub fn api_key(&self) -> Option<&str>

Returns the configured API key if any.

pub async fn chat( &self, request: ChatCompletionRequest, ) -> Result<ChatCompletionResponse, ClientError>

Sends a non-streaming chat completion request to /chat/completions.

pub async fn chat_stream( &self, request: ChatCompletionRequest, ) -> Result<ChatCompletionStream<impl Stream<Item = Result<Bytes, Error>> + Unpin>, ClientError>

Sends a streaming chat completion request to /chat/completions, returning an SSE chunk stream.

The request handshake awaits HTTP response headers. Once headers arrive and status is validated, the connection remains open for streaming tokens. Total request timeout is omitted to support long generations; connection timeout is governed by ClientBuilder::connect_timeout.

Stalls and Inactivity Timeouts

Because total request timeouts are omitted during streaming to permit long generations, a network disruption or hung server connection mid-stream may not terminate automatically. Callers in production environments should wrap stream consumption in an inactivity or idle timeout (for example, using tokio::time::timeout between successive chunk yields) so that mid-stream stalls can be detected and recovered from.

pub async fn embeddings( &self, request: EmbeddingRequest, ) -> Result<EmbeddingResponse, ClientError>

Generates vector embeddings for input text via /embeddings.

pub async fn models(&self) -> Result<ListModelsResponse, ClientError>

Lists models available from the provider via /models.

Trait Implementations

impl Clone for Client

fn clone(&self) -> Client

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Client

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

Formats the value using the given formatter.