Letta is an open-source framework designed for building stateful, long-term memory-enabled LLM applications. It allows developers to create advanced reasoning agents with transparent memory management, supporting various LLM backends like OpenAI, Anthropic, and Ollama. Letta offers a white-box, model-agnostic approach, enabling users to deploy agents via Docker or pip, with PostgreSQL as the recommended database for persistence. The framework includes the Letta Agent Development Environment (ADE), a graphical interface for managing and interacting with agents, and provides REST API and SDKs for integration. Letta is ideal for applications requiring persistent, intelligent agents, such as customer support chatbots, and supports both self-hosted and cloud-based deployments.