Skip to main content
MCP server CX Codex

LLM Gateway — Claude Code MCP server: what it does, how to install

LLM Gateway is an MCP server for Claude Code that routes prompts across model providers and tracks token spending and response latency inside the agent.

no license ★ 1,755 updated 2026-10-29

Install into Claude Code · Platform CLI: remote server All methods and platforms ↓
claude mcp add --transport http gateway https://api.llmgateway.io/mcp
Open on GitHub

What LLM Gateway does

LLM Gateway acts as an intermediary layer between your development environment and providers such as Anthropic, OpenAI, or Google Vertex AI. It normalizes requests using the OpenAI API standard, giving you a centralized spot to supervise token consumption, track response latency, and compare model costs.

You can link your tools to the hosted cloud endpoint at https://api.llmgateway.io/mcp or run your own instance via Docker. When deploying the unified image locally, mount named Docker volumes for PostgreSQL data to avoid container permission crashes during database setup.

What it helps with

  • route agent queries to multiple inference vendors from one place
  • audit token usage and spending across different models
  • monitor latency variations between providers
  • consolidate vendor API credentials into a single gateway

Example prompts for the agent

  • Check my total token usage and expenses across all models
  • Send this prompt to an Anthropic model via the gateway
  • Inspect response latency metrics for recent queries

How to install LLM Gateway

Platform CLI: remote server

claude mcp add --transport http gateway https://api.llmgateway.io/mcp

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about LLM Gateway

How to install LLM Gateway in Claude Code?

Run claude mcp add --transport http gateway https://api.llmgateway.io/mcp in your terminal.

What does LLM Gateway do?

It unifies communication with multiple AI providers through one API, logging query costs and tracking performance figures.

What is required when self-hosting with Docker?

You must assign Docker-managed named volumes for PostgreSQL storage rather than host directory bind mounts, preventing permission failures on /var/lib/postgresql/data.

Similar MCP servers

All MCP servers for AI and ML →

Project sponsors

Companies that keep this analytics open