Open WebUI

Open WebUI can connect to external model services through OpenAI-compatible configuration. When self-hosting Open WebUI, you can set the Tokeness Base URL and API key via environment variables or the admin panel.

Preparation

Field	Value
OpenAI API Base URL	https://n.tokeness.dev/v1
OpenAI API Key	A key created in the Tokeness console
Model	The full model name copied from the Tokeness model catalog

If your Open WebUI has been running for a while, some environment variables may have already been persisted to the database. When later changes do not take effect, update them in the admin panel or follow Open WebUI's persistent configuration guide.

Docker Environment Variable Method

For single-machine testing, start like this:

docker run -d \
  --name open-webui \
  -p 3000:8080 \
  -e OPENAI_API_BASE_URL=https://n.tokeness.dev/v1 \
  -e OPENAI_API_KEY=your Tokeness API key \
  -v open-webui:/app/backend/data \
  ghcr.io/open-webui/open-webui:main

If you use docker compose, you can write:

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - "3000:8080"
    environment:
      OPENAI_API_BASE_URL: https://n.tokeness.dev/v1
      OPENAI_API_KEY: ${TOKENESS_API_KEY}
    volumes:
      - open-webui:/app/backend/data

volumes:
  open-webui:

Put the key in a .env file:

TOKENESS_API_KEY=your Tokeness API key

Admin Panel Method

If Open WebUI is already running, you can configure it in the admin panel:

1. Log in to Open WebUI.
2. Open the Admin Panel.
3. Find the Connections or Models settings.
4. Locate the OpenAI API configuration.
5. Set the Base URL to https://n.tokeness.dev/v1.
6. Enter your Tokeness API key.
7. Save and refresh the page.

Menu names may differ across versions. As long as the field means OpenAI Base URL and OpenAI API Key, fill in the values above.

Set the Default Model

If you want new users to see a specific model by default, set:

DEFAULT_MODELS=YOUR_MODEL_NAME

Replace YOUR_MODEL_NAME with the full model name from the Tokeness model catalog.

For multiple models, separate with commas:

DEFAULT_MODELS=model-a,model-b

RAG and Embedding

Open WebUI's knowledge base, RAG, voice, and image features may have separate configuration items. Get standard chat working first, then configure these.

Common embedding fields:

RAG_OPENAI_API_BASE_URL=https://n.tokeness.dev/v1
RAG_OPENAI_API_KEY=your Tokeness API key

The embedding model name must be a vector model available in Tokeness. Do not put a chat model in the vector model field.

Verification

1. Open Open WebUI.
2. Select a model from Tokeness.
3. Send "reply with ok only".
4. Open the Tokeness usage logs and confirm a request arrived.
5. If you configured a knowledge base, upload a small text file and test retrieval.

Troubleshooting

Symptom	Action
Environment variable changes do not take effect	Open WebUI may have persisted the config — update it in the admin panel or handle PersistentConfig
Model list is empty	Enter the Tokeness model name manually
401	Check OPENAI_API_KEY
404	Check that OPENAI_API_BASE_URL is https://n.tokeness.dev/v1
Chat works but knowledge base does not	Check the embedding model and RAG-related configuration separately
Requests fail inside the Docker container	Enter the container and check network, DNS, and proxy settings

References

• Open WebUI OpenAI-Compatible Provider
• Open WebUI Environment Variables