Working Around the Compute Ceiling: Byte-Exact Memory in Galahad Makes LLM Reading a One-Time Cost LLM Reading a One-Time Cost
Galahad is a memory layer for LLMs that stores model data and key-value states to make reading text a one-time cost. It uses two memories, Taliesin for KV state reuse and Blaise for document section storage. This allows models to answer repeated questions about the same data without recomputing everything, shifting inference from computation to loading.