Which LLM memory for AI Agents?
Executive Summary<br>Project Breakdownsmem0ai/mem0 (⭐57.3k)<br>MemPalace/mempalace (⭐53.2k)<br>Lum1104/Understand-Anything (⭐47.8k)<br>pingcap/tidb (⭐40.1k)<br>volcengine/OpenViking (⭐25k)<br>supermemoryai/supermemory (⭐23.5k)<br>humanlayer/12-factor-agents (⭐22.9k)<br>rohitg00/agentmemory (⭐20.3k)<br>memvid/memvid (⭐15.6k)<br>vectorize-io/hindsight (⭐15.4k)
Cross-Cutting Analysis<br>Conflict Resolution Taxonomy<br>Recommendations<br>Executive Summary<br>The GitHub memory topic spans 6,187+ public repositories — a sprawling landscape that includes system memory profilers, AI agent memory layers, distributed databases, and knowledge graphs. The top 10, however, tell a more focused story: eight out of ten are AI agent memory projects , a category that barely existed two years ago and now dominates the conversation.<br>What the top 10 reveal:<br>AI Agent Memory (8 projects): mem0, MemPalace, Understand-Anything, OpenViking, supermemory, agentmemory, memvid, hindsight<br>Database/Infrastructure (1): TiDB — a distributed SQL database that has repositioned itself for agentic workloads<br>Principles/Framework (1): 12-Factor Agents — a methodology, not software<br>A fundamental architectural divide runs through the ecosystem: embedded/local-first projects (MemPalace, memvid, agentmemory) keep data and inference on-device, while client-server/cloud projects (mem0, supermemory, OpenViking, hindsight) rely on remote infrastructure. A small subset — supermemory on Cloudflare, mem0 on FastAPI+Postgres — leans fully into cloud-native architectures.<br>Project Breakdowns<br>1. mem0ai/mem0 — Universal Memory Layer for AI Agents
Aspect<br>Detail
Stars<br>⭐57,257
Language<br>Python (53%), TypeScript (42%)
License<br>Apache 2.0
Funding<br>YC S24, $24M raised
Latest<br>May 31, 2026; 326 releases
Website<br>https://mem0.ai
Overview. mem0 positions itself as a universal memory layer for AI agents, offering multi-level memory (User/Session/Agent), graph memory support, multi-signal retrieval (semantic, BM25, entity), and integrations with over 30 vector stores. It is the most well-funded project in the space, with Y Combinator backing and a $24M raise.<br>Architecture & dependencies. Built on Python 3.9+ with qdrant-client, pydantic, openai, and sqlalchemy at its core. The optional ecosystem is vast: 30+ vector stores (Chroma, Pinecone, Weaviate, Milvus, pgvector, FAISS), 24+ LLM providers, 15+ embedders, and 5 rerankers. Graph memory uses Neo4j 5.x. Self-hosted deployments require FastAPI, PostgreSQL, and Docker.<br>Strengths.<br>Top benchmark scores: 91.6 LoCoMo, 94.8 LongMemEval, 64.1 BEAM<br>Single-pass ADD-only algorithm avoids the complexity of in-place updates<br>Massive provider ecosystem with no single-vendor lock-in at the infrastructure level<br>Multi-signal retrieval combining entity linking with temporal reasoning<br>Rich surface area: MCP server, browser extension, CLI, Python and TypeScript SDKs<br>Peer-reviewed publication at ECAI 2025<br>Limitations.<br>Requires an external LLM (defaults to OpenAI, creating a de facto dependency)<br>Self-hosted setup is complex — Docker, PostgreSQL, and Neo4j are all prerequisites<br>The pre-April 2026 algorithm was significantly less capable<br>Deduplication only activates with infer=True, which is easy to miss<br>A known issue: silent memory loss when batch embedding partially fails<br>Graph memory adds meaningful overhead for marginal gain in some use cases<br>Conflict resolution approach. mem0's architecture is fundamentally ADD-only — memories accumulate, nothing is overwritten. Conflicts are resolved at retrieval time through multi-signal ranking (semantic similarity, BM25, entity matching, temporal recency). The old algorithm used a more traditional detection → recency evaluation → explicitness check → merge-or-replace → logging pipeline. Graph memory introduces LLM-driven entity/relation extraction with duplicate merging via semantic similarity. Deduplication uses a cosine-similarity threshold controlled by the infer flag.<br>2. MemPalace/mempalace — Local-First AI Memory System
Aspect<br>Detail
Stars<br>⭐53,198
Language<br>Python (94%)
License<br>MIT
Latest<br>v3.3.6 (May 24, 2026)
Website<br>https://mempalaceofficial.com
Overview. MemPalace is a local-first AI memory system inspired by the method of loci — a classical mnemonic technique. It stores content verbatim (never summarizes or lossy-compresses) and retrieves via semantic search. A knowledge graph with temporal validity, AAAK compression index, and an MCP server with 29 tools rounds out the offering.<br>Architecture & dependencies. Pure Python 3.9+ with ChromaDB 1.5+, huggingface_hub, and ONNX Runtime. The default multilingual embedding model is ~300 MB, with a 30 MB English-only alternative. The knowledge graph lives in SQLite. All embeddings run locally via ONNX — no API keys required.<br>Strengths.<br>Exceptional benchmark results: 96.6% R@5 raw, 98.4% hybrid, 99%+ with LLM reranking<br>Truly local-first: zero external API calls by default, no telemetry<br>Verbatim storage guarantee — never summarizes or applies lossy...