EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective

TL;DR

EvoMemBench introduces a comprehensive benchmark to evaluate various memory mechanisms in LLM agents, highlighting current limitations and guiding future improvements.

Contribution

The paper presents EvoMemBench, a systematic benchmark for agent memory, comparing 15 methods across different memory scopes and contents.

Findings

Long-context baselines are highly competitive.

Memory is most helpful when context is insufficient or tasks are difficult.

No single memory method outperforms others across all settings.

Abstract

Recent benchmarks for Large Language Model (LLM) agents mainly evaluate reasoning, planning, and execution. However, memory is also essential for agents, as it enables them to store, update, and retrieve information over time. This ability remains under-evaluated, largely because existing benchmarks do not provide a systematic way to assess memory mechanisms. In this paper, we study agent memory from a self-evolving perspective and introduce EvoMemBench, a unified benchmark organized along two axes: memory scope (in-episode vs. cross-episode) and memory content (knowledge-oriented vs. execution-oriented). We compare 15 representative memory methods with strong long-context baselines under a standardized protocol. Results show that current memory systems are still far from a general solution: long-context baselines remain highly competitive, memory helps most when the current context is insufficient or tasks are difficult, and no single memory form works consistently across all settings. Retrieval-based methods remain strong for knowledge-intensive settings, whereas procedural and long-term memory methods are more effective for execution-oriented tasks when their stored experience matches the task structure. We hope EvoMemBench facilitates future research on more effective memory systems for LLM-based agents. Our code is available at https://github.com/DSAIL-Memory/EvoMemBench.