Show HN: Semantic routing을 사용하여 LLM 추론 GPU 호출 94% 절감

Show HN: I reduced LLM inference GPU calls by 94% using semantic routing