OpenAI

Setup

import { OpenAIProvider } from "noumen";

const provider = new OpenAIProvider({
  apiKey: "sk-...",
  model: "gpt-4o",           // default model
  baseURL: "https://...",     // optional, for compatible APIs
});

Options

Compatible APIs

The baseURL option lets you use any OpenAI-compatible API. This includes:

• Azure OpenAI -- point to your Azure endpoint
• OpenRouter -- use https://openrouter.ai/api/v1
• Local models -- use Ollama, vLLM, or other local servers

const provider = new OpenAIProvider({
  apiKey: "your-key",
  baseURL: "https://openrouter.ai/api/v1",
  model: "anthropic/claude-sonnet-4",
});

Streaming

The OpenAI provider uses stream: true and stream_options: { include_usage: true } to get token usage on the final chunk. No additional configuration is needed.

Models

Any model available through the OpenAI chat completions API works. Common choices:

• gpt-4o -- fast, capable, good default
• gpt-4.1 -- latest generation
• gpt-4o-mini -- cheaper, faster for simpler tasks
• o3-mini -- reasoning model