Field Notes

Building an Open-Source Semantic Cache for LLM Calls

GPTCache is unmaintained. Nothing production-ready exists. 30% of your API spend is redundant. Here's the architecture and why it matters.