Provider Setup
This section guides you through configuring various LLM providers in CCX.
General Workflow
- Log in to the CCX admin console (default
http://localhost:3000) - Select the appropriate proxy endpoint (Chat, Messages, etc.)
- Click "Add Channel"
- Fill in the channel configuration
- Save and test
Key Configuration Fields
| Field | Description |
|---|---|
| Name | Display name for the channel |
| Service Type | Upstream API protocol: openai, claude, gemini, responses |
| Base URL | Upstream API endpoint |
| API Keys | Authentication keys (supports multi-key rotation) |
| Supported Models | Model allowlist for this channel |
| Model Mapping | Map request model names to upstream model names |
| Priority | Lower number = higher priority |
Service Type Guide
| Provider | Chat Endpoint | Messages Endpoint | Notes |
|---|---|---|---|
| DeepSeek | openai / https://api.deepseek.com | claude / https://api.deepseek.com/anthropic | Both protocols |
| GLM (Zhipu) | openai / https://open.bigmodel.cn/api/paas/v4 | claude / https://open.bigmodel.cn/api/anthropic | Both protocols |
| MiniMax | openai / https://api.minimax.io/v1 | claude / https://api.minimax.io/anthropic | Both protocols |
| Kimi / Kimi Code | openai / https://api.moonshot.cn/v1 or https://api.kimi.com/coding/v1 | claude / https://api.kimi.com/coding/ | Pay-as-you-go and Kimi Code endpoints |
| OpenAI GPT | openai / https://api.openai.com/v1 | — | OpenAI only |
| Xiaomi MiMo | openai / https://api.siliconflow.cn/v1 | — | Via SiliconFlow |
| Claude | claude (translation) | claude / https://api.anthropic.com | Native Messages |
| Gemini | openai or gemini | — | OpenAI-compat and native |
TIP
Most Chinese LLM providers now support both OpenAI Chat and Anthropic Messages protocols. If you use Claude Code CLI, you can configure the Anthropic-compatible endpoint directly under the Messages endpoint.
If you want to configure by client, see the Client Setup guide.