Crate openai_chatgpt_api 

Typed async Rust client for the OpenAI API.

Covers the current (2026) API surface:

Module	Endpoint
responses	/v1/responses — primary generation API (text, tools, reasoning, streaming)
chat	/v1/chat/completions — Chat Completions (incl. streaming)
embeddings	/v1/embeddings
images	/v1/images/generations, /v1/images/edits
audio	/v1/audio/speech, /v1/audio/transcriptions, /v1/audio/translations
moderations	/v1/moderations
models	/v1/models

Quick start

use openai_chatgpt_api::OpenAiClient;
use openai_chatgpt_api::responses::ResponseRequest;

#[tokio::main]
async fn main() -> Result<(), openai_chatgpt_api::OpenAiError> {
    let client = OpenAiClient::from_env()?; // OPENAI_API_KEY
    let request = ResponseRequest::new("gpt-5.6-terra", "Explain ownership in Rust.")
        .instructions("You are a concise assistant.")
        .max_output_tokens(500);
    let response = client.responses().create(&request).await?;
    println!("{}", response.output_text());
    Ok(())
}

Streaming

use futures_util::StreamExt;
use openai_chatgpt_api::OpenAiClient;
use openai_chatgpt_api::responses::{ResponseRequest, ResponseStreamEvent};

let client = OpenAiClient::from_env()?;
let request = ResponseRequest::new("gpt-5.6-luna", "Write a haiku.");
let mut stream = client.responses().stream(&request).await?;
while let Some(event) = stream.next().await {
    if let ResponseStreamEvent::OutputTextDelta { delta, .. } = event? {
        print!("{delta}");
    }
}

Model names are passed as plain strings: the API’s model lineup changes faster than any type system should chase. See https://platform.openai.com/docs/models for the current list.

Modules

audio

Audio APIs: speech synthesis (/v1/audio/speech), transcription (/v1/audio/transcriptions), and translation (/v1/audio/translations).

chat

Chat Completions API (/v1/chat/completions).

embeddings

Embeddings API (/v1/embeddings).

images

Images API (/v1/images/generations, /v1/images/edits).

models

Models API (/v1/models).

moderations

Moderations API (/v1/moderations).

responses

Responses API (/v1/responses) — OpenAI’s primary generation API.

Structs

DeletedObject

Result of a DELETE call (responses, fine-tuned models).

FilePart

An in-memory file attached to a multipart request (audio and image endpoints).

OpenAiClient

Asynchronous OpenAI API client.

OpenAiClientBuilder

Builder for OpenAiClient.

Enums

OpenAiError

Errors returned by crate::OpenAiClient.

Constants

DEFAULT_BASE_URL

Default API endpoint.

Type Aliases

EventStream

A stream of parsed server-sent events returned by the streaming endpoints.