llm_configs:
  general_configs:
    max_token_length: 24576
    temperature: 1.0
    top_p: 1
    # ---- For API backends ----
    max_workers: 16
    max_retries: 5
    request_timeout: 120
  
  # ---- Local models served by vLLM ----
  # Replace the `null` path to your local model path
  qwen2.5-7b:
    backend: vllm
    model: "Qwen/Qwen2.5-7B-Instruct"
    model_path: null
  
  qwen3-8b:
    backend: vllm
    model: "Qwen/Qwen3-8B"
    model_path: null

  gemma-3-4b:
    backend: vllm
    model: "google/gemma-3-4b-it"
    model_path: null
  
  qwen3-4b:
    backend: vllm
    model: "Qwen/Qwen3-4B-Instruct-2507"
    model_path: null
  
  # ---- Remote models served by the OpenRouter API ----
  # Requires the OPENROUTER_API_KEY environment variable.
  llama-3.3-70b:
    backend: openrouter
    model: "meta-llama/llama-3.3-70b-instruct"
  
  gpt-4o-mini:
    backend: openrouter
    model: "openai/gpt-4o-mini"

embedding_configs:
  bge-m3:
    model: "BAAI/bge-m3"
    model_path: null