Model Configuration

model_list is the new model-centric config pattern. Define each LLM once with all its credentials, then reference it by name everywhere else in config.

ModelConfig fields

Field	Type	Description
model_name	string	User-defined alias (e.g. "claude")
model	string	Provider model ID (e.g. "claude-sonnet-4-6")
api_base	string	Custom API endpoint (optional)
api_key	SecureString	API key — stored in .security.yml, never in config.json
proxy	string	HTTP proxy URL (optional)
auth_method	string	Auth scheme override (optional)
connect_mode	string	Connection mode override (optional)
workspace	string	Workspace path override (optional)
rpm	int	Requests-per-minute rate limit (optional)
request_timeout	int	Timeout in seconds (optional)
thinking_level	string	Extended thinking: "none" | "low" | "high" (Claude only)

Example: ~/.khunquant/config.json

{
  "model_list": [
    {
      "model_name": "claude",
      "model": "claude-sonnet-4-6",
      "api_key": "ANTHROPIC_API_KEY"
    },
    {
      "model_name": "gpt4o",
      "model": "gpt-4o",
      "api_key": "OPENAI_API_KEY"
    },
    {
      "model_name": "local",
      "model": "llama3.2",
      "api_base": "http://localhost:11434/v1",
      "api_key": "ollama"
    }
  ]
}

API keys referenced above are resolved from .security.yml — the value in config.json is just the key name.

Example: ~/.khunquant/.security.yml

model_list:
  - model_name: claude
    api_key: sk-ant-...
  - model_name: gpt4o
    api_key: sk-...
  - model_name: local
    api_key: ollama

tip

Use enc:// prefix to store encrypted values in .security.yml. See .security.yml Reference for details.

Selecting the default model

Set the default model for your agent in the agent config:

{
  "agents": {
    "defaults": {
      "model_name": "claude"
    }
  }
}