Chatbox
Add an OpenAI-compatible provider to Chatbox
Chatbox supports custom providers for Chat Completions, Responses, and Claude. Prepare your gateway key and model.
1. Install Chatbox
Download the desktop app from chatboxai.app, install it, and launch it. Desktop supports direct API setup; the web version also depends on gateway CORS configuration.
2. Add a provider
Open Settings → Model Provider and choose Add provider:
| Field | Value |
|---|---|
| Name | VNAI Pro |
| API Mode | OpenAI API Compatible |
Click Add, then configure the new provider:
| Field | Value |
|---|---|
| API Key | A key created on your gateway website |
| API Host | https://api.vnaipro.app/v1 |
| API Path | /chat/completions |
| Model ID | An ID from Check models |
The Preview must show https://api.vnaipro.app/v1/chat/completions.
Do not put /v1/chat/completions in API Path when API Host already ends in /v1.
3. Select and test
Add the model ID to the provider's model list if missing, and save. Start a new conversation, choose VNAI Pro and your model, send a short prompt, and check usage history on your gateway website.
Other protocols
| Model/channel | API Mode | API Host | API Path |
|---|---|---|---|
| Responses | OpenAI Responses API Compatible | https://api.vnaipro.app/v1 | /responses |
| Native Claude | Claude API Compatible | https://api.vnaipro.app/v1 | /messages |
Create a separate provider for each protocol. Chatbox's Claude adapter appends /messages to the host,
so the host includes /v1, unlike Claude Code's root-URL setting.
If Responses mode is missing, update Chatbox or select a Chat Completions model.
Chat works but the knowledge base fails: embeddings need separate configuration; a chat key/model does not automatically provide embeddings. Connection error: check Preview and common errors.
Sources: Chatbox, API modes in the UI, URL handling.