Large Language Models (LLMs) have become increasingly popular, transforming a wide range of applications across various domains. However, the real-world effectiveness of their query cache systems has not been thoroughly investigated. In this work, we for the first time conducted an analysis on real-world human-to-LLM interaction data, identifying key challenges in existing caching solutions for LLM-based chat services. Our findings reveal that current caching methods fail to leverage semantic connections, leading to inefficient cache performance and extra token costs. To address these issues, we propose SCALM, a new cache architecture that emphasizes semantic analysis and identifies significant cache entries and patterns. We also detail the implementations of the corresponding cache storage and eviction strategies. Our evaluations show that SCALM increases cache hit ratios and reduces operational costs for LLMChat services. Compared with other state-of-the-art solutions in GPTCache, SCALM shows, on average, a relative increase of 63% in cache hit ratio and a relative improvement of 77% in tokens savings.

Aman's background is in the intersection of Business Applications and Artificial Intelligence, using both to drive the next generation of business applications Aman also founded and worked in various startups in search, social, trading systems, and enterprise software. His last startup was TopCorner, a political platform for micro-lobbying. Aman was the architect for IBM SuperSell Enterprise and Oracle CRM. He was previously the Director of Special Projects for the CEO's office at Oracle. Aman earned a MS in Computer Science with research focused on natural language processing (NLP) from Stanford.