Open WebUI

Get going in 30 seconds

Create a docker-compose.yml with:

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - "3000:8080"
    environment:
      OPENAI_API_BASE_URL: https://northerninference.ca/v1
      OPENAI_API_KEY: ni_live_YOUR_KEY_HERE
    volumes:
      - open-webui:/app/backend/data

Then:

docker compose up -d

Open http://localhost:3000, sign in, go to Settings → Admin Settings → Connections, and click Refresh to pull the model list. Pick any NI route and start chatting.

Open WebUI is a self-hostable ChatGPT-style frontend. It supports OpenAI-compatible backends natively.

Docker compose snippet

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - "3000:8080"
    environment:
      OPENAI_API_BASE_URL: https://northerninference.ca/v1
      OPENAI_API_KEY: ni_live_YOUR_KEY_HERE
      # Optional: enable multiple endpoints (e.g. NI + local Ollama)
      # OPENAI_API_BASE_URLS: "https://northerninference.ca/v1;http://ollama:11434/v1"
      # OPENAI_API_KEYS: "ni_live_YOUR_KEY;ollama"
    volumes:
      - open-webui:/app/backend/data

docker compose up -d

Open http://localhost:3000, sign in, go to Settings → Admin Settings → Connections. The NI connection should be live; Refresh to pull the model list.

Per-user NI keys (multi-tenant)

Open WebUI has a Settings → Connections tab for each user. If you host Open WebUI internally and each user has their own NI key, let them paste it there instead of setting a global key in the compose file. NI will bill each user's team separately.

Route selection

The selected model ID is the route contract. Copy exact routing keys from /models or /api/billing/models into Open WebUI's model list. NI API key allowed route tiers control which routes the key may call.