MemGPT

About

LLM operating system with virtual memory. Lets LLMs manage their own memory for long-running conversations and tasks.

• Virtual memory management with automatic data paging between working and external storage
• Built-in support for OpenAI, Azure OpenAI, and local LLMs via vLLM or llama.cpp
• REST API and Python SDK for integrating MemGPT into custom applications and workflows
• Tool-use framework enabling the LLM to call external functions (e.g., search, calculator, file I/O)
• Memory editing and inspection via CLI or web interface to manually correct or view stored data