hedging8563/tokenlab-mcp-server

What it does

TokenLab AI gateway MCP server acts as a unified interface for accessing multiple large language models including OpenAI, Anthropic, and Google Gemini. It provides model and pricing discovery capabilities, allowing users to explore available models and their costs without requiring API keys for catalog operations. The server supports OpenAI-compatible Chat Completions, native Anthropic Messages, and Gemini inference endpoints.

Key features

• Multi-model support: OpenAI, Anthropic, and Google Gemini inference
• Pricing discovery and model catalog browsing without authentication
• OpenAI-compatible Chat Completions API
• Native protocol support for Anthropic Messages and Gemini
• Optional API key configuration for inference operations
• MCP (Model Context Protocol) server implementation

How to set up

Install the MCP server from the GitHub repository. For catalog operations, no API key is required. To use inference features, optionally configure your API keys for OpenAI, Anthropic, and/or Google Gemini. The server integrates with Claude, ChatGPT, and other MCP-compatible applications through standard MCP protocol configuration.