STORY · FORSKNING_
Zero-Mem: Memory without token cost for LLM agents
Researchers present Zero-Mem, a method that lets LLM agents handle long-term memory without additional LLM calls or tokens. The system organizes interaction history in an entity-context graph and temporal hierarchy, retrieving relevant context before generating final responses. The method achieves competitive performance and reduces time spent on memory operations by 57.6 percent.
WHY IT MATTERS
This improves the efficiency of LLM agents by eliminating hidden token costs in memory management, which has practical significance for scaling agent systems that need to handle long dialogue histories.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.