Letta
Framework and cloud for building stateful AI agents with long-term memory, from the creators of MemGPT.
Overview
Letta is an open-source (Apache-2.0) framework for building stateful agents that manage their own long-term memory, commercializing the MemGPT research on self-editing LLM memory. Spun out of UC Berkeley's Sky Lab, it emerged from stealth in September 2024 with a $10M seed led by Felicis. Letta agents persist state through APIs so they can learn across interactions rather than resetting each session. It is offered as an open-source server plus Letta Cloud, with backing from angels including Jeff Dean and Clem Delangue.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 24 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
ASEAN Perspective
Letta in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
What this is for: Building agents that keep persistent, self-managed memory and state across long-running interactions.
Who this is for: Developers who need agents with durable memory rather than stateless prompt-and-response loops.
Availability: Free open-source framework plus a paid Letta Cloud; based on the MemGPT research from UC Berkeley.
What people say
Letta comes from a serious research background. It's the commercial spin-out of MemGPT, the 2023 UC Berkeley paper (1,000+ citations in under a year) that framed agent memory as an OS-style hierarchy paging between in-context "RAM" and archival "disk." That lineage, a $10M Felicis seed, and angels like Jeff Dean and Clem Delangue give it credibility, and the open-source repo has grown to roughly 23K stars — a leading option for memory-centric agents.
Practitioner comparisons draw a clear line rather than a rave. The recurring comparison is that Mem0 is a lightweight memory layer you attach to any framework. Letta is a full agent runtime: more powerful and self-managing, but also heavier and more opinionated. You're adopting a platform, not a drop-in library. Reviewers warn against picking on stars alone and urge teams to evaluate its persistence model, temporal accuracy, latency, and compliance. The implicit caution is that the OS-memory abstraction adds overhead, both conceptually and operationally.
It's an ambitious take on stateful agents and well-regarded by those who need durable memory. But the commitment to a full runtime means it isn't the easy bolt-on some expect.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Letta's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Letta unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Letta directly →
Spotted something out of date? Suggest an update →
Alternatives to Letta
More in Agents & Automation
