Skip to main content
Lukan supports multiple model providers. All providers implement a common Provider interface returning AsyncIterable<StreamEvent>.

Available Providers

Setting Up a Provider

Interactive Setup

This interactive wizard helps you select a provider and enter your API key.

Manual Configuration

  1. Set the API key via environment variable:
  2. Or add it to ~/.config/lukan/credentials.json:
Credentials are read from environment variables first, then from ~/.config/lukan/credentials.json.

Selecting Models

You can also specify the provider and model directly:

Per-Channel Configuration

You can override the provider per-channel (WhatsApp, Email) in config.json:

OAuth Authentication

GitHub Copilot

This initiates an OAuth device flow to authenticate with GitHub.

OpenAI Codex

This initiates OAuth authentication with OpenAI Codex.

Ollama Cloud

Ollama Cloud fetches models dynamically from https://ollama.com/api/tags. Vision-capable models are detected automatically (patterns: vision, -vl, multimodal, llava, llama-4).

OpenAI Compatible

Generic provider for any OpenAI-compatible endpoint (vLLM, Ollama local, LM Studio, etc.). Requires a base URL:
Models are fetched dynamically from the /models endpoint. The API key is optional.

Lukan Cloud

Lukan Cloud is coming soon. All models, one API key, usage dashboard — no need to manage multiple provider keys.