Catalog Systems Recent method papers — theorized, no distinct product LightMem

LightMem

Lightweight, efficient memory-augmented generation.

Active · Verified 2026-10-28 · fresh as of 2026-10-28 · 25 cited facts, 1 with a third-party source · Suggest a correction

At a glance

Type: Lightweight memory-augmented vendor fix
Tier: T4
Section: Recent method papers — theorized, no distinct product
Created: 2025-06 vendor fix
Latest release: no releases fix
License: MIT repo fix
GitHub: 820★ +30/mo Python repo fix
Pricing: not applicable — research paper fix
Funding: not applicable — not commercial fix
Freshness: 2026-10-28 fix
Layer: agent-infra repo fix
Licensing: open-source repo fix
Developer activity: OpenRank 0.5 (12-mo avg 3.67) · 5 participants — 2026-09 third-party fix

Memory taxonomy

storage: vector
retrieval: similarity
persistence: long-term
update: extraction
unit: fact
governance: inspectable
conflict: none

When to use

Optimised for: not applicable - research paper

Anti-fit: not applicable - research paper

Pros & cons

Pros

Lightweight memory store for resource-constrained agents.

Cons

Preprint; capability ceiling lower than full memory layers.

Claims & capabilities

Sensory/short-term/long-term human-cognition memory system with offline sleep-time consolidation. Headline: QA accuracy +2.09-6.40% (GPT) and up to +7.67% (Qwen) while reducing token usage 38x/21.8x and API calls 30x/55.5x; baselines: LangMem, A-MEM, MemoryOS, Mem0, NaiveRAG, FullText; primary datasets: LongMemEval (LongMemEval-S) and LoCoMo.

Technical surface

API surface: not applicable — research paper
Backend storage: not applicable — research paper
Deployment: not applicable — research paper
Embedding model: not applicable — research paper
Multi-tenancy: not applicable — research paper
MCP: not applicable — research paper, no deployed product
A2A: not applicable — research paper, no deployed product
OpenTelemetry: not applicable — research paper, no deployed product

Similar systems

Other recent method papers — theorized, no distinct product in the catalog, ranked by inbound references.

Compressive Transformer T3
Maintains recent states in full resolution while compressing older memories with learned compression functions. DeepMind.
MemGPT v2 / agent-tools T3
Already in catalog as the foundational MemGPT paper. Note: Letta is the productionised successor (cross-listed).
Transformer-XL T3
Extends context through segment-level recurrence + caching of hidden states from prior segments. Foundational long-context architecture.
Generative Agents T3
Park et al. — landmark agent-simulation paper. Reflection + memory stream + retrieval enable believable agent behavior.
MemoryBank T3
Enhances LLMs with long-term memory. Early influential paper.
Reflexion T3
Language agents with verbal reinforcement learning.

Sources (24 cited facts, 1 third-party)

Each fact links to the page it came from; the badge says who published it (how to read sources).

What it is Lightweight memory-augmented vendor
What it does & what's distinct Lightweight, efficient memory-augmented generation. vendor
Claims / benefits Sensory/short-term/long-term human-cognition memory system with offline sleep-time conso… vendor
Modalities text vendor
Created 2025-06 vendor
License MIT repo
GitHub 820★ +30/mo Python repo
Mindshare 49 cites · 7.6/mo · 822* / 74 forks · 16 influential SS cites paper
Citations 48 cites 48/yr paper
Founders / pedigree no data — searched arxiv.org/abs/2510.18866; author affiliations not found in preview vendor
Memory volume / scale no scale claim — research vendor
Performance +2.09–6.40% LongMemEval-S/LoCoMo (GPT) 38x token reduction vendor
Reproducibility Original team + downstream attention: 74 forks of https://github.com/zjunlp/LightMem; pu… repo
Code/weights release Code public + weights closed — https://github.com/zjunlp/LightMem · 822* / 74 forks · la… repo
Schema evolution schema with migration vendor
Contradiction handling LLM resolves and overwrites vendor
Pros Lightweight memory store for resource-constrained agents. vendor
Cons Preprint; capability ceiling lower than full memory layers. vendor
Project & sources arxiv 2510.18866 GitHub zjunlp/LightMem vendor
commit-trajectory [{"ym":"2025-06","cum":1},{"ym":"2025-08","cum":18},{"ym":"2025-09","cum":26},{"ym":"202… repo
citation-trajectory [{"year":2025,"cum":3,"infl":0},{"year":2026,"cum":4,"infl":1}] paper
licensing-bucket open-source repo
taxonomy-three-layer-rollup agent-infra repo
developer-activity-opendigger {"activity":8.43,"as_of":"2026-09","openrank_12mo":3.67,"openrank_current":0.5,"particip… third-party