feat: recommend models for inference by auto-detecting host hardware specifications

Integrate llmfit or similar functionality for local LLM inference