LLM Kit OpenAI

OpenAI provider for LLM Kit - Complete integration with OpenAI's chat completion API.

Note: This provider uses the standardized builder pattern. See the Quick Start section for the recommended usage.

Features

• Text Generation: Generate text using GPT models with support for all OpenAI chat models
• Streaming: Stream responses in real-time
• Tool Calling: Support for function calling
• Multi-modal: Support for text, images, audio, and PDFs
• Reasoning Models: Special handling for o1, o3, and other reasoning models
• Provider Options: Logprobs, reasoning effort, service tiers, and more
• Type-safe Configuration: Builder pattern for easy setup

Installation

Add this to your Cargo.toml:

[dependencies]
llm-kit-openai = "0.1"
llm-kit-core = "0.1"
llm-kit-provider = "0.1"
tokio = { version = "1", features = ["full"] }

Quick Start

Using the Client Builder (Recommended)

use llm_kit_openai::OpenAIClient;
use llm_kit_provider::language_model::LanguageModel;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create provider using the client builder
    let provider = OpenAIClient::new()
        .api_key("your-api-key")  // Or use OPENAI_API_KEY env var
        .build();
    
    // Create a language model
    let model = provider.chat("gpt-4o");
    
    println!("Model: {}", model.model_id());
    println!("Provider: {}", model.provider());
    Ok(())
}

Using Settings Directly (Alternative)

use llm_kit_openai::{OpenAIProvider, OpenAIProviderSettings};
use llm_kit_provider::language_model::LanguageModel;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Create provider with settings
    let provider = OpenAIProvider::new(OpenAIProviderSettings::default());
    
    let model = provider.chat("gpt-4o");
    
    println!("Model: {}", model.model_id());
    Ok(())
}

Configuration

Environment Variables

Set your OpenAI API key as an environment variable:

export OPENAI_API_KEY=your-api-key
export OPENAI_BASE_URL=https://api.openai.com/v1  # Optional

Using the Client Builder

use llm_kit_openai::OpenAIClient;

let provider = OpenAIClient::new()
    .api_key("your-api-key")
    .base_url("https://api.openai.com/v1")
    .organization("org-123")
    .project("proj-456")
    .header("Custom-Header", "value")
    .name("my-openai-provider")
    .build();

Using Settings Directly

use llm_kit_openai::{OpenAIProvider, OpenAIProviderSettings};

let settings = OpenAIProviderSettings::new()
    .with_api_key("your-api-key")
    .with_base_url("https://api.openai.com/v1")
    .with_organization("org-123")
    .with_project("proj-456")
    .add_header("Custom-Header", "value")
    .with_name("my-openai-provider");

let provider = OpenAIProvider::new(settings);

Builder Methods

The OpenAIClient builder supports:

• .api_key(key) - Set the API key
• .base_url(url) - Set custom base URL
• .organization(org) - Set OpenAI organization ID
• .project(project) - Set OpenAI project ID
• .name(name) - Set provider name
• .header(key, value) - Add a single custom header
• .headers(map) - Add multiple custom headers
• .build() - Build the provider

Supported Models

All OpenAI chat models are supported, including:

GPT-4 Family

• gpt-4 - Most capable GPT-4 model
• gpt-4-turbo - Faster GPT-4 variant
• gpt-4o - Optimized GPT-4 model
• gpt-4o-mini - Smaller, faster GPT-4o variant

GPT-3.5 Family

• gpt-3.5-turbo - Fast and efficient model

Reasoning Models

• o1 - Latest reasoning model
• o1-preview - Preview version of o1
• o1-mini - Smaller o1 variant
• o3-mini - Next-generation reasoning model

GPT-5 Family

• Future models will be supported as they become available

For a complete list of available models, see the OpenAI Models documentation.

OpenAI-Specific Features

Reasoning Models

OpenAI reasoning models (o1, o1-preview, o1-mini, o3-mini) have special handling:

• Developer role: System messages automatically use the "developer" role instead of "system"
• Parameter filtering: Unsupported settings (temperature, top_p, presence_penalty, frequency_penalty, etc.) are automatically removed
• Token limits: Uses max_completion_tokens instead of max_tokens

These adjustments happen automatically when you use a reasoning model, so you don't need to make any code changes.

Provider-Specific Options

OpenAI supports additional options beyond the standard LLM Kit parameters:

Reasoning Effort

Control the computational effort for reasoning models:

use llm_kit_openai::chat::{OpenAIChatLanguageModelOptions, openai_chat_options::ReasoningEffort};

let options = OpenAIChatLanguageModelOptions {
    reasoning_effort: Some(ReasoningEffort::High),
    ..Default::default()
};

Available values:

• ReasoningEffort::Low - Faster, less thorough reasoning
• ReasoningEffort::Medium - Balanced reasoning
• ReasoningEffort::High - More thorough, slower reasoning

Logprobs

Request log probabilities for generated tokens:

use llm_kit_openai::chat::{OpenAIChatLanguageModelOptions, openai_chat_options::LogprobsOption};

let options = OpenAIChatLanguageModelOptions {
    logprobs: Some(LogprobsOption::Number(5)),  // Top 5 token probabilities
    ..Default::default()
};

Service Tier

Select the service tier for processing:

use llm_kit_openai::chat::{OpenAIChatLanguageModelOptions, openai_chat_options::ServiceTier};

let options = OpenAIChatLanguageModelOptions {
    service_tier: Some(ServiceTier::Auto),
    ..Default::default()
};

Available values:

• ServiceTier::Auto - Automatic tier selection
• ServiceTier::Default - Standard processing tier

Organization and Project

Configure organization and project IDs:

let provider = OpenAIClient::new()
    .api_key("your-api-key")
    .organization("org-123")
    .project("proj-456")
    .build();

Usage Examples

Basic Text Generation

See examples/chat.rs for a complete example.

Streaming Responses

See examples/stream.rs for a complete example.

Tool Calling

OpenAI supports function calling for tool integration. See examples/chat_tool_calling.rs for a complete example.

Examples

See the examples/ directory for complete examples:

• chat.rs - Basic chat completion
• stream.rs - Streaming responses
• chat_tool_calling.rs - Tool calling with chat models
• stream_tool_calling.rs - Streaming with tool calling

Run examples with:

cargo run --example chat
cargo run --example stream
cargo run --example chat_tool_calling
cargo run --example stream_tool_calling

Documentation

• API Documentation
• LLM Kit Documentation
• OpenAI API Reference

License

Licensed under:

• MIT license (LICENSE-MIT)

Contributing

Contributions are welcome! Please see the Contributing Guide for more details.