scripts/config/multi_model_config.json

{
  "servers": [
    {
      "name": "research_fast",
      "url": "http://127.0.0.1:1234/v1",
      "model": "qwen/qwen3-4b-2507",
      "timeout": 120.0,
      "max_concurrent": 2,
      "weight": 1.0,
      "enabled": true,
      "description": "Fast Qwen 4B model for research tasks - quick exploration and data gathering"
    },
    {
      "name": "thinking_analysis", 
      "url": "http://127.0.0.1:1234/v1",
      "model": "qwen/qwen3-4b-thinking-2507",
      "timeout": 150.0,
      "max_concurrent": 2,
      "weight": 1.2,
      "enabled": true,
      "description": "Qwen Thinking model for analysis and critique - reasoning focused"
    },
    {
      "name": "synthesis_quality",
      "url": "http://127.0.0.1:1234/v1", 
      "model": "google/gemma-3-12b",
      "timeout": 180.0,
      "max_concurrent": 1,
      "weight": 1.5,
      "enabled": true,
      "description": "Gemma 12B model for synthesis - high-quality final output generation"
    }
  ],
  "agent_mapping": {
    "research": "research_fast",
    "analysis": "thinking_analysis", 
    "synthesis": "synthesis_quality",
    "critic": "thinking_analysis"
  },
  "load_balance_strategy": "agent_type_mapping",
  "health_check_interval": 30.0,
  "failover_enabled": true,
  "debug_mode": false,
  "notes": {
    "setup": "Single LM Studio server with multiple models",
    "models_required": [
      "qwen/qwen3-4b-2507",
      "qwen/qwen3-4b-thinking-2507", 
      "google/gemma-3-12b"
    ],
    "expected_behavior": "Agents use different models but same server endpoint",
    "performance": "Parallel requests with model switching handled by LM Studio"
  }
}