Open-source platform for building AI agents that remember, learn, and self-improve across conversations over time.
Letta (formerly MemGPT) is an open-source framework for building stateful AI agents with persistent long-term memory. It gives LLMs the ability to remember past interactions, learn new skills, and improve through experience. Agents manage their own context windows using a tiered memory system inspired by computer architecture. Letta is model-agnostic, ships with Python and TypeScript SDKs, and offers both self-hosted and cloud deployment options. It is built for developers creating assistants, support bots, coding agents, and any application where AI needs to retain knowledge across sessions.
Letta is the evolved platform built on the original MemGPT research project. MemGPT introduced the concept of treating an LLM's context window like virtual memory in an operating system. Letta takes that core idea and wraps it in a full production framework with APIs, SDKs, a visual development environment, and cloud hosting. The MemGPT package has been renamed to Letta.
Yes. Letta offers a free plan with 5,000 monthly credits, 2 agent templates, and 1 GB of storage. The open-source server can also be self-hosted at no cost under the Apache 2.0 license. Paid plans start at $20/month (Pro) for higher limits and additional features.
Letta is model-agnostic and supports OpenAI, Anthropic, vLLM, Ollama, and other LLM API backends. That said, performance is best with top-tier commercial models. Local and open-source models may produce less reliable results.
Yes. The Letta server is fully open source under Apache 2.0. You can self-host it on your own infrastructure using Docker, giving you complete control over data and deployment. A managed cloud option is also available for teams that prefer not to run their own servers.
Letta uses a three-tier memory architecture. Core memory lives in the LLM's context window for immediate access. Recall memory stores searchable conversation history outside the context window. Archival memory provides long-term storage that agents query through tool calls. Agents manage their own memory by deciding what to save, update, or discard.
Letta Code is a memory-first coding agent available as a desktop app for macOS, Windows, and Linux. It uses git-backed memory to learn your codebase and coding patterns over time. It supports skills, subagents, and works across any model provider. It ranked #1 among model-agnostic open-source agents on the Terminal-Bench coding benchmark.
Letta provides official SDKs for Python and TypeScript, plus a full REST API. You can integrate Letta agents into any application that can make HTTP requests, regardless of your tech stack.
Letta offers deeper, more structured memory with white-box control, making it better for long-term maintainability and specialized applications. Mem0 is simpler to set up as a drop-in memory layer. Zep focuses on research-driven memory sophistication. Choose Letta if you want full open-source control and are willing to invest in configuration.
0 out of 5 stars
Based on 0 reviews
5 star reviews
4 star reviews
3 star reviews
2 star reviews
1 star reviews
If you've used this tool, share your thoughts with other users
Build stateful AI agents with persistent memory
AI-native Creative Platform
AI that does your manual work across 1,500+ apps
AI writing and knowledge assistant for teams
AI notepad that turns meeting notes into polished docs.
Effortless PDF editing, signing, and collaboration.
Observability that installs itself and fixes bugs
Visual website builder with AI and no-code power
AI work platform for teams and projects