Developer Resources

LibreChat

Configure a LibreChat custom endpoint for RunAPI, mount its YAML in Docker, restart, and verify a chat.

LibreChat custom endpoints are deployment configuration, not a per-user connection form. You need access to the LibreChat deployment files and a standard RunAPI API key from the Authentication Guide.

Install LibreChat

Install LibreChat with the official local setup or use your existing deployment. Complete the installation before adding the custom endpoint.

Add the RunAPI endpoint

Create or update librechat.yaml in the LibreChat deployment directory. Replace YOUR_RUNAPI_MODEL_ID with a current identifier from the Model Catalog.

YAML

version: 1.3.13
endpoints:
  custom:
    - name: "RunAPI"
      apiKey: "${RUNAPI_API_KEY}"
      baseURL: "https://runapi.ai/v1"
      models:
        default: ["YOUR_RUNAPI_MODEL_ID"]
        fetch: true
      titleConvo: true
      titleModel: "current_model"
      modelDisplayLabel: "RunAPI"

Store the key in .env, not directly in the YAML file:

SHELL

RUNAPI_API_KEY=YOUR_RUNAPI_API_TOKEN

For Docker Compose, make sure docker-compose.override.yml mounts the configuration file:

YAML

services:
  api:
    volumes:
      - ./librechat.yaml:/app/librechat.yaml

Restart the LibreChat deployment after changing any of these files.

Select a model

models.default guarantees at least one known model. With models.fetch: true, LibreChat also requests the current model inventory. Keep a valid default identifier even when discovery is enabled so the endpoint remains usable if discovery is temporarily unavailable.

Verify the connection

After restart, select RunAPI from the endpoint selector, choose the configured model, and send Reply with the word connected. A normal reply confirms both the deployment configuration and the model request.

Troubleshoot

RunAPI is absent from the endpoint selector: validate the YAML indentation, confirm docker-compose.override.yml mounts the file, and restart the API container.
The endpoint loads but authentication fails: confirm .env defines RUNAPI_API_KEY for the API container and recreate the container if its environment is stale.
No fetched models appear: keep one current identifier in models.default, confirm baseURL ends in /v1, and inspect the API container logs.
LibreChat ignores edited YAML: verify the host path and container path in the volume mount point to the file you changed.

For the shared protocol behavior, use the LLM API Quickstart. For request and response fields, use the Chat Completions API Reference.