Chatbox
Chatbox supports custom OpenAI-API-compatible providers. Fill in the host and key, then add the models you want.
Add the provider
- Open Settings → Model Provider and click Add at the bottom.
- Enter a name and choose OpenAI API Compatible as API Mode.
- Fill in the key and address as in the table.
- Check that the Preview line under the fields shows the full endpoint (below the table).
| Field | Value |
|---|---|
| API Mode | OpenAI API Compatible |
| API Key | Your NezhaGate API key |
| API Host | https://nezhagate.com/v1 |
| API Path | Leave empty |
Preview should read https://nezhagate.com/v1/chat/completions.
Add models
Click Fetch to load the model list, or New to add one by hand (such as gpt-5.5). Add chat models only, tick the capabilities each one has (image input, reasoning, tool use) and set its context window; models left unconfigured are treated as text-only.
Test
Click Check: Chatbox sends a few short tests to the chosen model (text, image, tools) and shows Connection successful. Image and tool tests failing on a text-only model do not matter.
The web app
The Chatbox web app calls the API straight from the browser, and the NezhaGate API does not accept cross-origin browser requests. Switch on Improve Network Compatibility in the provider settings (requests then pass through Chatbox servers), or use the desktop or mobile app.
Claude (optional)
Add another provider with Claude API Compatible as API Mode and https://nezhagate.com/anthropic/v1 as API Host.
Troubleshooting
The host already ends in /v1, so the path must stay empty; otherwise the address becomes /v1/v1/chat/completions.
The host must go up to /v1: https://nezhagate.com/v1.
Check that the key was copied in full without extra spaces and that the balance is enough; on the web app, also switch on Improve Network Compatibility.
