RepoRadar
Leave a Tip
Item detail

MemTensor/MemOS

MemTensor/MemOS is an Apache-2.0 'memory operating system' for LLMs and AI agents that ships a self-evolving memory layer with hybrid retrieval, scheduled memory management, and an arXiv-backed design so long-running agents and chat apps can keep persistent, structured memory across sessions and platforms.

Open source →Back to radar
Score7.9
Popularity7.6
Riskconditional
TierGold
Score breakdown
Usefulness8.0
Novelty8.0
Momentum8.0
Maturity6.4
Open-source/build8.4
Evidence7.2
Workflow potential9.0
Setup ease6.4

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

Useful for agent builders, research teams, and product engineers who need a structured, persistent memory layer for long-running LLM apps and AI agents that survives across sessions, providers, and platforms without bolting together a vector DB and a hand-written memory manager.

Who should use it

agent builders who need a structured, persistent memory layer for long-running LLM appsresearch teams looking for an arXiv-backed, open-source memory design they can extendproduct engineers building cross-session, cross-platform AI assistants that need to remember contextteams that want a hybrid-retrieval memory surface without a hosted-memory vendor lock-in

Who should skip it

Skip if the source link, docs, or setup requirements do not match your workflow.

Risk explanation

It persists structured memory across sessions and platforms, so confirm what gets stored, where the storage lives, who can read it, and what the default retention and redaction settings are before connecting a production agent or chat app.

Evidence links

Closest alternatives / related signals

memoryagentragretrievalllmopen-source