Local-First LLM

Routes LLM requests to a local model (Ollama, LM Studio, llamafile) before falling back to cloud APIs. Tracks token savings and cost avoidance in a persisten...