neuron-provider-openai

OpenAI provider for the neuron agent blocks ecosystem. Implements the Provider and EmbeddingProvider traits from neuron-types against the OpenAI Chat Completions and Embeddings APIs, supporting synchronous completions, server-sent event (SSE) streaming, and text embeddings.

The default completion model is gpt-4o. The default embedding model is text-embedding-3-small. The default base URL is https://api.openai.com. All can be overridden with the builder API. The base_url override also makes this client usable with Azure OpenAI and compatible third-party endpoints.

Key Types

OpenAi -- client struct with builder methods (new, model, base_url, organization). Implements Provider and EmbeddingProvider from neuron-types.
ProviderError -- re-exported error type for provider failures.
EmbeddingError -- re-exported error type for embedding failures.
EmbeddingRequest, EmbeddingResponse -- re-exported embedding types.
StreamHandle -- returned by complete_stream, yields StreamEvent items as the model generates tokens.

Features

Full message mapping: text, tool calls, tool results, images.
Organization header support for multi-org OpenAI accounts.
ToolChoice support: Auto, Any, Required, Specific(name).
SSE streaming parsed from raw byte stream with data: [DONE] sentinel handling.
Usage statistics included in streaming responses via stream_options.
EmbeddingProvider support for generating text embeddings with optional dimension control.

Usage

Completions

use neuron_provider_openai::OpenAi;
use neuron_types::{CompletionRequest, ContentBlock, Message, Provider, Role, SystemPrompt};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let provider = OpenAi::new("sk-...")
        .model("gpt-4o")
        .organization("org-...");

    let request = CompletionRequest {
        messages: vec![Message {
            role: Role::User,
            content: vec![ContentBlock::Text("Explain Rust's ownership model.".into())],
        }],
        system: Some(SystemPrompt::Text("You are a helpful assistant.".into())),
        max_tokens: Some(1024),
        ..Default::default()
    };

    let response = provider.complete(request).await?;

    for block in &response.message.content {
        println!("{block:?}");
    }
    Ok(())
}

Embeddings

use neuron_provider_openai::OpenAi;
use neuron_types::{EmbeddingProvider, EmbeddingRequest};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let provider = OpenAi::new("sk-...");

    let request = EmbeddingRequest {
        input: vec!["Hello world".to_string(), "Rust is great".to_string()],
        dimensions: Some(256),
        ..Default::default()
    };

    let response = provider.embed(request).await?;

    println!("Model: {}", response.model);
    println!("Embeddings: {} vectors", response.embeddings.len());
    println!("Dimensions: {}", response.embeddings[0].len());
    println!("Usage: {} tokens", response.usage.total_tokens);
    Ok(())
}

Part of neuron

This crate is one block in the neuron composable agent toolkit. It depends only on neuron-types.

License

Licensed under either of Apache License, Version 2.0 or MIT License at your option.

neuron-provider-openai 0.3.0