AI SDK OpenAI - Rust Implementation

OpenAI provider implementation for the AI SDK in Rust. This crate provides a unified interface for interacting with OpenAI's models including GPT, DALL-E, Whisper, and Embeddings.

Features

• 🤖 Chat Completion - GPT-4, GPT-3.5 and other chat models
• 🔢 Embeddings - text-embedding-3-small, text-embedding-3-large
• 🖼️ Image Generation - DALL-E 2 and DALL-E 3
• 🗣️ Speech Synthesis - Text-to-speech with OpenAI TTS
• 👂 Transcription - Speech-to-text with Whisper
• 🔄 Streaming - Stream responses for real-time applications
• 🛠️ Tool Calling - Function calling support for chat models

Installation

Add this to your Cargo.toml:

[dependencies]
ai-sdk-openai = "0.1"
ai-sdk-provider = "0.1"

Quick Start

Chat Completion

use ai_sdk_openai::OpenAIChatModel;
use ai_sdk_provider::{LanguageModel, CallOptions, Message};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let api_key = std::env::var("OPENAI_API_KEY")?;
    let model = OpenAIChatModel::new("gpt-4", api_key);

    let options = CallOptions {
        messages: vec![
            Message::user("What is Rust?"),
        ],
        ..Default::default()
    };

    let response = model.do_generate(options).await?;
    println!("Response: {}", response.text);

    Ok(())
}

Embeddings

use ai_sdk_openai::OpenAIEmbeddingModel;
use ai_sdk_provider::EmbeddingModel;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let api_key = std::env::var("OPENAI_API_KEY")?;
    let model = OpenAIEmbeddingModel::new("text-embedding-3-small", api_key);

    let embeddings = model.do_embed(vec!["Hello, world!".to_string()]).await?;
    println!("Embedding dimension: {}", embeddings[0].len());

    Ok(())
}

Image Generation

use ai_sdk_openai::OpenAIImageModel;
use ai_sdk_provider::{ImageModel, ImageOptions};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let api_key = std::env::var("OPENAI_API_KEY")?;
    let model = OpenAIImageModel::new("dall-e-3", api_key);

    let options = ImageOptions {
        prompt: "A beautiful sunset over mountains".to_string(),
        ..Default::default()
    };

    let result = model.do_generate(options).await?;
    println!("Generated image URL: {}", result.url);

    Ok(())
}

Supported Models

Chat Models

• gpt-4 - Most capable GPT-4 model
• gpt-4-turbo - GPT-4 Turbo with improved performance
• gpt-3.5-turbo - Fast and efficient model

Embedding Models

• text-embedding-3-small - Smaller, faster embedding model
• text-embedding-3-large - Larger, more accurate embedding model

Image Models

• dall-e-2 - DALL-E 2 image generation
• dall-e-3 - DALL-E 3 with improved quality

Speech Models

• tts-1 - Standard text-to-speech
• tts-1-hd - High-definition text-to-speech
• whisper-1 - Speech-to-text transcription

Documentation

• API Documentation
• Examples
• AI SDK Provider Traits

Repository

This crate is part of the AI SDK for Rust workspace.

Contributing

We welcome contributions! Please see the Contributing Guide for guidelines.

License

Licensed under either of:

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.