OpenMem
To build world-leading open-source infrastructure for large-model-oriented memory, providing AI systems with continuous growth and context retention capabilities.
At a Glance
- AI Developers
- Enterprise AI Teams
- Cloud Service Providers
AI Tools by OpenMem
(1)
MemOS
AI Agent Memory Framework
Discussions
No discussions yet
Be the first to start a discussion about OpenMem
Latest News
MemOS OpenClaw plugin (Cloud & Local) launch
MemOS 2.0 'Stardust' Official Release
MemTensor wins Gold Medal at Ascend AI Innovation Contest 2025
LightMem: Lightweight and Efficient Memory-Augmented Generation paper published
Products & Services
A memory-native framework and operating system for AI agents and LLMs, providing persistent, low-latency memory infrastructure.
A high-performance universal text embedding model ranked #1 on the MTEB benchmark.
A lightweight pluggable memory system for LLMs using semantic compression and noise filtering.
A unified memory operation language that converts natural language into standardized JSON commands for memory management.
Market Position
Pioneer in 'Memory-Native' infrastructure for LLMs, offering a comprehensive open-source OS (MemOS) that bridges the gap between static knowledge and dynamic task-oriented memory.
Leadership
Founders
Feiyu Xiong
Founder and CEO of MemTensor. Previously a Principal Researcher at the Institute for Advanced Algorithms Research (IAAR) and an AI technical expert at Alibaba.
Zhiyu Li
CTO of MemTensor. Previously an AI technical expert at Alibaba and researcher at Renmin University of China.
Bo Tang
Core member and researcher at MemTensor. Researcher at the Institute for Advanced Algorithms Research (IAAR).
Executive Team
Feiyu Xiong
Founder & CEO
Former AI expert at Alibaba, researcher at IAAR.
Zhiyu Li
CTO
Former AI expert at Alibaba, researcher at IAAR/RUC.
Board of Directors
Founding Story
MemTensor was incubated by the Shanghai Algorithm Innovation Research Institute (IAAR) to solve the memory constraints of LLMs and agents through a native memory operating system (MemOS).
Business Model
Revenue Model
Subscription-based model for memory APIs and knowledge base services, along with enterprise custom deployments and industry solutions.
Pricing Tiers
50K add, 20K search, 3M/1M tokens per month, 10 KB items.
600K add, 200K search, 12M/4M tokens, 30 KB items. Original price $286/month.
Unlimited resources, private deployment, dedicated support.
Target Markets
- AI Developers
- Enterprise AI Teams
- Cloud Service Providers
- AI Companions and personalized assistants
- Autonomous AI Agents with long-term memory
- Enterprise knowledge bases and customer support
- Gaming, education, and healthcare AI applications
- China Telecom
- Alibaba Cloud
- Huawei Cloud