Chatbox¶
Add the gateway to Chatbox as a custom provider and your chats are checked against your policies, charged to your team and recorded in Request logs. One provider lists every model your key may call, from every provider.
What it is¶
Chatbox is a chat app for desktop, mobile and the web. Each person runs their own copy, and it calls models directly from that device. Its custom providers speak the OpenAI, Anthropic or Gemini format.
The two settings¶
Open Settings › Model Provider, choose Add provider, give it a Name (for example "Vidai") and choose the API Mode OpenAI API Compatible. Then:
| Field | Value |
|---|---|
| API Host | https://<your-gateway>/v1 |
| API Path | /chat/completions (the default; leave it) |
| API Key | Your Vidai key (starts vk_), never a provider's key |
| Models | Fetch to list the models your key may call, or add names by hand |
The Preview under the fields should read https://<your-gateway>/v1/chat/completions.
The gateway's address is under Connect your app on the Keys page and in System › Settings › Gateway. You get your key from Keys › Issue a key; its secret is shown once.
The other modes work too, if you prefer them:
| API Mode | API Host | Chatbox calls |
|---|---|---|
| Claude API Compatible | https://<your-gateway>/v1 |
/v1/messages |
| Google Gemini API Compatible | https://<your-gateway> |
/v1beta/models/… |
| OpenAI Responses API Compatible | https://<your-gateway>/v1 |
/v1/responses: OpenAI and Azure OpenAI models only |
The gateway translates Messages and Gemini calls too, so each of the first two can reach GPT, Claude and Gemini models.
A minimal call¶
- Save the provider; Check sends a test call.
- Start a new chat and pick a model under your provider, for example
claude-sonnet-4.5. - Send "Reply with exactly: ok". A short answer means the address and key are right.
Streaming¶
Chatbox streams answers. A stream from another provider's model arrives in the format of the mode you chose; Vidai records the tokens and the cost either way.
Tool calls¶
Chatbox's tools (web search, MCP servers, knowledge base) travel as tool calls and work on every provider the gateway translates to. Mark a model as able to use tools in its settings if Chatbox doesn't know it.
Reading what the gateway did¶
Chatbox doesn't show response headers. Look the call up in Request logs: filter on Key, Model or Path. Cost is never in the answer.
Errors¶
Chatbox shows the gateway's error message in the chat. Common ones: 401 invalid_api_key, 403 model_not_allowed,
403 content_guardrail_blocked (a guardrail stopped the message), 429 (a limit on your key or team). The full list is
in Headers and errors.
Model names¶
Fetch calls GET /v1/models, which the gateway answers with exactly the models your key may call, named as your
organisation approved them under Models in the Control room.
What goes through the gateway¶
| Chatbox feature | Through the gateway |
|---|---|
| Chats, streaming, images in messages, tools, conversation names, with your gateway provider selected | Yes, to OpenAI, Azure OpenAI, Anthropic, Gemini and Vertex AI models |
| Other providers in Chatbox, with their own keys | No |
Check it works in a minute¶
Send "Reply with exactly: ok" to a gateway model, then open Request logs a few seconds later and filter on your key.
Good to know¶
- Chatbox is one person's app, so each person uses their own key, issued in Keys › Issue a key; their calls show under them and are charged to their team.
- Improve Network Compatibility isn't needed for the gateway; leave it off.
- Chatbox's own docs: Help centre.