LibreChat through Model Gate

Add a custom endpoint to librechat.yaml:

version: 1.2.1

endpoints:
  custom:
    - name: "Model Gate"
      apiKey: "${MODEL_GATE_API_KEY}"
      baseURL: "https://api.model-gate.com/v1"
      models:
        default:
          - "gpt-5.4"
          - "ch-47"
          - "ch-46s"
        fetch: true
      titleConvo: true
      titleModel: "ch-46s"
      modelDisplayLabel: "Model Gate"

Set the secret before starting LibreChat:

export MODEL_GATE_API_KEY="mg_live_..."

Restart LibreChat after changing the configuration. When fetch: true is supported by the installed LibreChat version, it reads canonical model IDs from the minimal /v1/models list. That list no longer includes aliases, capability flags, limits or key-specific prices; configure any required extra metadata from the authenticated Models and prices page.