Integration Guide

RunPod + Meridian

Run your own models on RunPod's GPU cloud for privacy and control. Use Meridian for frontier models when you need maximum capability. One API surface, zero lock-in.

1

Deploy a model on RunPod

Spin up a GPU pod with your model of choice — Llama, Mistral, or any GGUF/Ollama-compatible model. Note the pod ID and port.

2

Expose an OpenAI-compatible endpoint

Use RunPod's built-in OpenAI-compatible server or deploy vLLM / text-generation-inference. Meridian speaks the OpenAI chat completions protocol.

3

Add your endpoint to Meridian

In your Meridian dashboard, navigate to Settings → Custom Providers. Paste your RunPod endpoint URL and optional API key. Meridian will validate the connection.

4

Route requests

Use model routing rules to send specific models to RunPod. Everything else falls back to Meridian's frontier models automatically.