Recommended configuration
First verify in the terminal:
curl https://www.aifast.club/v1/models \
-H "Authorization: Bearer $API_KEY"Then add the OpenAI Compatible connection in the Connections area of the OpenWebUI management settings:
| Field | Recommended value |
|---|---|
| URL / Base URL | https://www.aifast.club/v1 |
| API Key | Independent Key created by AIFast console |
| model | from /v1/models Select from the returned results |
Different versions of OpenWebUI may place the entry in "Administrator Settings/Connection", "Settings/Model" or environment variables. To determine whether the configuration is correct, the final request address and connection test results should prevail.
Verification steps
- Refresh the model list after saving the connection.
- Choose a real model ID.
- Create a new blank session and just send "reply only OK".
- After validating common responses, then test streaming output, files, tools, and multimodal capabilities.
- Confirm in the server log that the request goes to the expected API.
Model list is empty
- direct request
/v1/models, confirm that the Key and network are available. - Check whether the OpenWebUI container can access the external domain name.
- Check for reverse proxy, DNS, and certificate errors.
- Confirm that the connection address is not appended repeatedly
/v1。
Return HTML instead of JSON
Typically indicates that the request went to a website page rather than an API route. Check that the final URL contains the correct /v1 path.
Streaming output appears once
Use first curl -N Request directly. If the terminal can be displayed in chunks but OpenWebUI is displayed in one go, focus on checking the reverse proxy buffer, container gateway and OpenWebUI version.
Security advice
- Create a separate API Key for OpenWebUI.
- Do not write keys to public Docker Compose, screenshots, or Git repositories.
- Set quotas and concurrency limits for multi-user environments.
- Regularly check usage logs and rotate abnormal keys.