Connect Open WebUI

Use an administrator account on your existing Open WebUI instance. This setup uses its server-side OpenAI-compatible connection and preserves other connections.

1. Add an API connection

Open Settings → Admin → Connections. Under Manage OpenAI API Connections, select + Add Connection.

Setting	Value
URL	The service Base URL above, including /v1
API Key	Your service key, entered privately without Bearer
Provider, under Advanced	Default
Model IDs	Leave empty for discovery, or allow only exact IDs you want

This uses OpenAI-compatible Chat Completions. The URL is an API base, so do not add /chat/completions or /models/. Default keeps the connection on the standard protocol without enabling server-specific model management.

Select Verify Connection to query /models/, then Save. Saving alone does not test the endpoint. Keep the connection’s enable switch on.

2. Choose a model

Start a new chat and select a text model from the added connection. For a manual Model IDs allowlist, copy exact IDs from the authenticated models endpoint. A model enabled by an administrator may still be restricted by your instance’s existing user permissions.

Send one short message and confirm the request in your service dashboard. Chat messages spend balance; automatic titles and other background tasks can generate additional requests. Review their model settings before broader use.

3. Keep the connection current

The admin UI saves the connection persistently. Refresh the model picker after changes. For 404, check the /v1 suffix; for missing models, check the allowlist, connection switch and key permissions.

This guide covers text chat. It does not configure embeddings, speech, images or browser-based Direct Connections, which have separate requirements. Agent tools also depend on model capabilities and the permissions you grant in Open WebUI.

See the official OpenAI-compatible connection guide and provider connection overview.