Skip to main content
The server has two provider IDs:
  • codex
  • claude
Both providers implement the same ChatProvider interface: stream, complete, and optional aiModel access for tool-using agent turns.

​
Codex

The codex provider uses the OpenAI AI SDK.
SettingValue
Required keyOPENAI_API_KEY
Default modelgpt-5.4
Modelsgpt-5.4, gpt-5.4-mini, gpt-5.4-nano

​
Claude

The claude provider uses the Anthropic AI SDK.
SettingValue
Required keyANTHROPIC_API_KEY
Default modelclaude-sonnet-4-6
Modelsclaude-sonnet-4-6, claude-opus-4-7, claude-haiku-4-5

​
Provider Info

GET /providers returns the models and whether each provider is configured: 
{
  "providers": [
    {
      "id": "codex",
      "label": "Codex",
      "defaultModel": "gpt-5.4",
      "models": ["gpt-5.4", "gpt-5.4-mini", "gpt-5.4-nano"],
      "configured": true
    }
  ]
}

​
Selection

HTTP requests choose a provider in the request body. 
{
  "provider": "codex",
  "model": "gpt-5.4-mini",
  "messages": [{ "role": "user", "content": "Explain this file." }]
}
WebSocket sessions can choose provider and model during session.create. If omitted, the server picks the first available provider.