inference-cost

A deterministic memory layer for local LLMs at near-zero cost

An open-source deterministic memory layer bolts onto frozen LLMs, enabling zero-forgetting recall beyond context at near-zero marginal cost.