Connect your app
Most clients with custom OpenAI-compatible providers need an address, an API key and a model.
Prepare these three values
| Setting | Value |
|---|---|
| Full API Base URL | https://guangjiapi.top/v1 |
| API Key | A model API key created in the Guangji API console |
| Model ID | Copy it from the marketplace for the same group as your key |
Changing only the address, or using a key from another provider, will not authenticate with Guangji API.
Which address should I enter?
Different clients build request paths differently. Check the resulting request URL:
| What the client expects | Enter |
|---|---|
| Full OpenAI Base URL | https://guangjiapi.top/v1 |
Root URL, with /v1 added automatically | https://guangjiapi.top |
| Complete chat-completions URL | https://guangjiapi.top/v1/chat/completions |
Chat-completions requests should reach /v1/chat/completions. A path containing /v1/v1/ adds the prefix twice.
Do not put /dashboard, /sign-in or a complete endpoint into a field that only accepts a base URL.
General setup
- Add a provider and choose OpenAI-compatible or its equivalent.
- Name it Guangji API and enter the appropriate address.
- Paste the complete API key.
- Fetch models or manually add an available model ID.
- Save, select that provider's model in your app, and send a short test request.
Choose your integration
- Desktop chat: Cherry Studio.
- Desktop AI assistant: WorkBuddy illustrated setup.
- Coding-tool configuration: CC Switch illustrated one-click import.
- Programs and services: API examples.
- Incremental responses: streaming.
Not all models support the same protocols or capabilities. Images, audio, tools and long context require support from the model, platform and client.