aiqbench analysis
This arXiv preprint proposes a memory system for LLM agents that rewrites dialogue into self-contained units, quantizes them into a learnable codebook, and adds co-occurrence-based retrieval alongside standard vector search. The reported gains are benchmark results on LongMemEval-S and LoCoMo with two backbones, plus faster memory construction than several existing baselines. Readers should treat these as unreplicated claims from a single paper, but the core question of how agents retrieve logically related yet lexically distant context is a practical one for anyone building long-running assistants.
MemCode: Discrete Semantic Representations for Long-Term Agent Memory
arXiv:2610.19472v1 Announce Type: new Abstract: Long-term memory remains a fundamental bottleneck for large language model (LLM) agents across sessions. Existing systems typically adopt a write-then-retrieve paradigm, storing context-dependent dialogue fragments and retrieving them via nearest-neighbor search in a continuous embedding space. This design yields unstable access units and systematically misses logically related but lexically distant evidence. We propose MemCode, a long-term agent memory framework that couples retrieval-oriented writing with discrete associative retrieval. MemCode rewrites dialogue into self-contained atomic memory units (AMUs), maps AMUs into a learnable codebook with isomorphic semantic quantization (ISQ), and uses codebook co-occurrence topology as a retrieval path complementary to vector search. For long-term memory stores, MemCode separates online atomic writes from asynchronous offline consolidation. On LongMemEval-S and LoCoMo with GPT-4o-mini and Qwen3-30B-A3B-Instruct-2507 backbones, MemCode achieves the best overall accuracy while keeping per-dialogue memory construction within roughly 40~130 seconds, making it 3~21 times faster than LightMem and over an order of magnitude faster than A-MEM and MemoryOS.
Abstract:Long-term memory remains a fundamental bottleneck for large language model (LLM) agents across sessions. Existing systems typically adopt a write-then-retrieve paradigm, storing context-dependent dialogue fragments and retrieving them via nearest-neighbor search in a continuous embedding space. This design yields unstable access units and systematically misses logically related but lexically distant evidence. We propose MemCode, a long-term agent memory framework that couples retrieval-oriented writing with discrete associative retrieval. MemCode rewrites dialogue into self-contained atomic memory units (AMUs), maps AMUs into a learnable codebook with isomorphic semantic quantization (ISQ), and uses codebook co-occurrence topology as a retrieval path complementary to vector search. For long-term memory stores, MemCode separates online atomic writes from asynchronous offline consolidation. On LongMemEval-S and LoCoMo with GPT-4o-mini and Qwen3-30B-A3B-Instruct-2507 backbones, MemCode achieves the best overall accuracy while keeping per-dialogue memory construction within roughly 40~130 seconds, making it 3~21 times faster than LightMem and over an order of magnitude faster than A-MEM and MemoryOS.
Artificial Intelligence (cs.AI)
arXiv:2610.19472 [cs.AI]
(or arXiv:2610.19472v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.19472 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 249 words
Related coverage
Presentation: From Models to Agents: Building Context-Aware Consumer AI at Scale at DoorDash
InfoQ · Aug 15, 2026
Beyond Similarity: Trustworthy Memory Search for Personal AI Agents
arXiv · Jun 6, 2026
WFM: Wiki Foundation Model for Complex Agentic Reasoning
arXiv · Sep 17, 2026
What is contextual retrieval? How AI agents find the right context
Neo4j · Sep 30, 2026