Documentation
Complete guide to LLM Link - a universal proxy for multiple LLM providers with hot-reload configuration and multi-protocol support.
Key Features
Multi-Provider Support
Connect to 10+ LLM providers including OpenAI, Anthropic, Zhipu, Volcengine, and more through a single unified interface.
Protocol Flexibility
Expose OpenAI, Ollama, or Anthropic-compatible APIs. Switch protocols without changing your client code.
Hot Configuration Reload
Update API keys, switch providers, and modify settings via REST API without service restart.
App Presets
First-class presets for Zed.dev, Codex CLI, and Claude Code with optimized ports and protocols.
Installation
Install from crates.io:
Configuration
Set API keys via environment variables:
Usage by Applications & Protocols
Zed.dev
Runs an Ollama-compatible endpoint for Zed on port 11434:
In Zed: Settings → AI → Custom provider → http://localhost:11434
Codex CLI
OpenAI-compatible endpoint for Codex on port 8088:
Configure Codex to use http://localhost:8088/v1 as OpenAI endpoint.
Claude Code
Anthropic-compatible endpoint for Claude Code on port 8089:
Point Claude Code base URL to http://localhost:8089 and set a dummy API token.
API Endpoints
GET /api/health Service health and version
GET /api/providers List configured providers
GET /api/models Available models
POST /api/config/update Update configuration
Troubleshooting
Port already in use
Use --port flag to specify a different port
API key not working
Verify key format and use hot-reload API to update
Connection timeout
Check network and provider base URL configuration