This seems to be another over optimization for AI that many are trying to get into. The LLM's improve, and your setup is deprecated, you wasted time optimizing for a slight edge. TDLR: You trade time for slight edge.