Struct OpenAI 

pub struct OpenAI { /* private fields */ }

The main OpenAI client.

See OpenAI API docs for the full API reference.

Use with_options() to create a cheap clone with per-request customization (extra headers, query params, timeout):

use openai_oxide::RequestOptions;
use std::time::Duration;

let custom = client.with_options(
    RequestOptions::new()
        .header("X-Custom", "value")
        .timeout(Duration::from_secs(30))
);

Implementations

impl OpenAI

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

Create a new client with the given API key.

pub fn with_config<C: Config + 'static>(config: C) -> Self

Create a client from a full config.

pub fn with_options(&self, options: RequestOptions) -> Self

Create a cheap clone of this client with additional request options.

The returned client shares the same HTTP connection pool (reqwest::Client uses Arc internally) but applies the merged options to every request.

use openai_oxide::RequestOptions;

let custom = client.with_options(
    RequestOptions::new().header("X-Custom", "value")
);
// All requests through `custom` will include the X-Custom header.
let resp = custom.chat().completions().create(req).await?;

pub fn client(&self) -> &Self

Returns a reference to self for direct access to low-level request methods.

Useful for napi/FFI bindings that need post_json_bytes etc.

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

Create a client using the OPENAI_API_KEY environment variable.

pub fn azure(config: AzureConfig) -> Result<Self, OpenAIError>

Create a client configured for Azure OpenAI.

Examples

use openai_oxide::{OpenAI, AzureConfig};

let client = OpenAI::azure(
    AzureConfig::new()
        .azure_endpoint("https://my-resource.openai.azure.com")
        .azure_deployment("gpt-4")
        .api_key("my-azure-key")
)?;

pub fn batches(&self) -> Batches<'_>

Access the Batches resource.

pub fn uploads(&self) -> Uploads<'_>

Access the Uploads resource.

pub fn beta(&self) -> Beta<'_>

Access the Beta resources (Assistants, Threads, Runs, Vector Stores).

pub fn audio(&self) -> Audio<'_>

Access the Audio resource.

pub fn chat(&self) -> Chat<'_>

Access the Chat resource.

pub fn models(&self) -> Models<'_>

Access the Models resource.

pub fn fine_tuning(&self) -> FineTuning<'_>

Access the Fine-tuning resource.

pub fn files(&self) -> Files<'_>

Access the Files resource.

pub fn images(&self) -> Images<'_>

Access the Images resource.

pub fn moderations(&self) -> Moderations<'_>

Access the Moderations resource.

pub fn responses(&self) -> Responses<'_>

Access the Responses resource.

pub fn embeddings(&self) -> Embeddings<'_>

Access the Embeddings resource.

pub fn conversations(&self) -> Conversations<'_>

Access conversation endpoints (multi-turn server-side state).

pub fn videos(&self) -> Videos<'_>

Access video generation endpoints (Sora).

pub fn realtime(&self) -> Realtime<'_>

Access the GA Realtime endpoints (post-2026 client_secrets API).

The legacy beta accessor (client.beta().realtime()) targets the retired /realtime/sessions and /realtime/transcription_sessions paths. Use this instead for new code.

pub async fn post_json_bytes( &self, path: &str, json_bytes: Bytes, ) -> Result<Value, OpenAIError>

Send a POST request with a pre-serialized JSON body and return raw JSON.

Skips serde_json::to_vec — the caller provides already-serialized bytes. Used by napi bindings where JS already has the JSON string, avoiding double serialization (napi copy + serde serialize).

pub async fn post_stream_json_bytes( &self, path: &str, json_bytes: Bytes, ) -> Result<Response, OpenAIError>

Send a POST request with pre-serialized JSON body and return an SSE stream.

Trait Implementations

impl Clone for OpenAI

fn clone(&self) -> OpenAI

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for OpenAI

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

Formats the value using the given formatter.