Using Downsizing with Open WebUI

Prerequisites

A running Open WebUI instance (self-hosted via Docker/Docker Compose, typically alongside Ollama) and a Downsizing account that can make requests (pay-per-use or a connected subscription both work). Grab your Inference ID and API key from the Inference API page.

Open WebUI is a self-hosted server — like LibreChat, it usually runs in its own container, separate from wherever you use the dsCLI. There's no local file on your machine for dsto write to, so wiring it up is a one-time connection you add either in the Admin Settings UI (stored in Open WebUI's own database) or via environment variables on the container.

Option A: Admin Settings UI

This stores the connection in Open WebUI's own database, so it survives container restarts without touching your deployment's environment.

1. Sign in to Open WebUI as an admin and go to ⚙️ Admin Settings → Connections.
2. Under OpenAI, click ➕ Add Connection.
3. Set the URL field to your Downsizing OpenAI-compatible endpoint, and the Key field to your Downsizing API key (starts with dws_):

http://localhost:3001/YOUR_INFERENCE_ID/openai

Replace YOUR_INFERENCE_IDwith your workspace UUID from the dashboard. If Downsizing doesn't auto-populate the model list via /models, add the model IDs you want to the Model IDs allowlist field, then click Save.

Option B: Environment variables

If you'd rather configure this at deploy time (so it's reproducible across restarts and redeploys), set Open WebUI's OPENAI_API_BASE_URL and OPENAI_API_KEY environment variables when starting the container:

OPENAI_API_BASE_URL=http://localhost:3001/YOUR_INFERENCE_ID/openai
OPENAI_API_KEY=dws_your_api_key_here

Or, in docker-compose.yml:

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    environment:
      - OPENAI_API_BASE_URL=http://localhost:3001/YOUR_INFERENCE_ID/openai
      - OPENAI_API_KEY=dws_your_api_key_here
      - ENABLE_OLLAMA_API=false

Restart the container (docker compose up -d) for the variables to take effect. Note that environment-variable connections and Admin Settings connections are independent — if you configure both, you'll see two OpenAI connections in the UI.

Verify

Start a new chat in Open WebUI, pick one of the Downsizing-routed models from the model picker, and send a test message. If you get a response, the connection is wired up correctly. Then check your savings dashboard to see the request appear in your usage metrics.