Updated:
Endpoints
Cubence provides the API routes below. Choose an endpoint for your region and network, then configure your client's Base URL for the protocol it uses.
Available Routes
| Route | API Endpoint |
|---|---|
| Mainland China — load balancing optimized for the three major networks | https://api-lb.cubence.ai |
| Global | https://api.cubence.ai |
| Outside Mainland China | https://api.cubence.com |
Base URL Configuration
The examples below use the Mainland China route. To use another route, replace the host and keep any /v1 suffix or request path required by your client.
| Protocol | BASE_URL | Clients |
|---|---|---|
| Anthropic / Gemini | https://api-lb.cubence.ai | Claude Code, Gemini CLI |
| OpenAI-compatible | https://api-lb.cubence.ai/v1 | Codex, Grok Build, Cherry Studio |
See the setup guide for your client for configuration examples. API request paths, such as /v1/images/generations, stay the same.
Configuration Tips
- Update existing configurations: Replace the old API host with your chosen endpoint from the table above, keeping any
/v1suffix or request path required by your client. - Check the protocol: Claude Code and Gemini CLI use the root URL. Codex and other OpenAI-compatible clients require
/v1in the Base URL. - When using a proxy: Add your chosen endpoint's hostname to your proxy whitelist, or disable the proxy for a direct connection.