Open WebUI

Deploy Open WebUI and manage a server-side gateway connection as administrator.

Install and prepare

Install Docker, then follow Open WebUI Quick Start or run the command below. Keep the named data volume so accounts and settings survive container replacement. Open http://localhost:3000 and create the initial administrator account. Create the connection key in routerra.ai Keys; obtain a permitted model ID from the gateway model list.

Configure the connection

As administrator, open Settings → Admin → Connections → Manage OpenAI API Connections → Add Connection. Set URL to https://routerra.ai/v1 and API Key to YOUR_API_KEY; use the default OpenAI-compatible connection type and enable it. Save. Models normally come from this connection’s /models endpoint. If discovery is unavailable, enter YOUR_MODEL_ID in its Model IDs allowlist. This is a server connection shared under your instance’s access rules, not a personal browser Direct Connection.

docker run -d --name open-webui -p 3000:8080 \
  -v open-webui:/app/backend/data \
  --restart always ghcr.io/open-webui/open-webui:main

Select a model and resolve errors

Select YOUR_MODEL_ID in a new chat and send a short message. If the model is absent, check the connection toggle and allowlist, then save to refresh the list. Existing database settings can override OPENAI_API_BASE_URL and OPENAI_API_KEY on restart: edit the Admin connection directly. ENABLE_PERSISTENT_CONFIG=False forces environment values but makes Admin edits nonpersistent. For network errors, inspect docker logs open-webui and verify outbound HTTPS from the container.