Prerequisites
- Access to a Dify console (cloud or self-hosted)
- A Qhaigc API key — get one from the API Tokens page
- An installed OpenAI-compatible model provider plugin in your Dify instance
Configuration Steps
Open model provider settings
In the Dify console, navigate to Settings and find the model provider or model settings section. Locate the entry point for managing model provider plugins.
Add an OpenAI-compatible provider plugin
Install or select the OpenAI model provider plugin that supports custom API endpoints. Dify’s plugin architecture means the exact entry point name may vary by version — look for a model provider plugin rather than a legacy menu item.
Enter your Qhaigc credentials
Fill in the credentials the plugin requires:
| Field | Value |
|---|---|
| API Key | Your Qhaigc API key (starts with sk-) |
| API Base URL / Endpoint | https://api.qhaigc.net/v1 |
Only include
/v1 in the base URL if the plugin field explicitly asks for it. Follow whatever label the plugin uses for the endpoint field.Fetch or configure models
If the plugin supports remote model fetching (
fetch-from-remote mode), sync the model list. If you need to add models manually, enter each model ID as required by the plugin’s interface.Select the model in your application
In your Dify chat app, Agent, or workflow, open the model selector and choose the provider and model you just configured.
Verifying the Connection
Your integration is working when:- The model provider page saves without errors
- The model is selectable in your Dify application
- A test conversation returns a valid response
Frequently Asked Questions
I can’t find an “OpenAI Compatible” entry — where do I look? Dify’s current architecture is plugin-based, not the older flat “model providers” page. Look for a model provider plugin in the plugin marketplace or your installed plugins list. The entry point name varies by Dify version. Why doesn’t the tutorial list specific model names? The models available depend on your provider plugin configuration and Dify version. Rather than list models that may not apply to your setup, configure the plugin and then fetch or enter models as needed. Do I have to include/v1 in the base URL?
Only if the plugin field or Dify’s documentation explicitly requires it. Follow the field label in the UI rather than assuming a fixed format.