Letta / MemGPT alternatives

Originated as MemGPT (UC Berkeley), where the model pages its own memory between context, recall and archival tiers. The product now centres on Letta Code, a stateful agent harness whose agents rewrite their own memory, skills and prompts over time. Full Letta / MemGPT record →

  1. 1 Mem0 Dedicated memory layers

    also in Dedicated memory layers · same storage (kv, vector) · same retrieval (similarity) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also open source · also self-hostable

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 67k released 2026-09-25 Apache-2.0 Compare with Letta / MemGPT →

  2. 2 Hindsight (Vectorize) Dedicated memory layers

    also in Dedicated memory layers · same storage (vector) · same retrieval (similarity) · same memory unit (episode, summary) · also open source · also self-hostable · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 45k released 2026-10-21 MIT Compare with Letta / MemGPT →

  3. 3 CrewAI Memory Framework-embedded memory

    same storage (kv, vector) · same retrieval (similarity, injection) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also open source · also self-hostable · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 51k released 2026-10-14 MIT Compare with Letta / MemGPT →

  4. 4 VEKTOR Memory Dedicated memory layers

    also in Dedicated memory layers · same storage (vector) · same retrieval (similarity) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also self-hostable · also offered managed

    OSS — Self-host ✓ Managed ✓ MCP ✓ ★ 5 released 2026-07-16 proprietary (commercial licence) Compare with Letta / MemGPT →

  5. 5 AGiXT Adaptive Memory Framework-embedded memory

    same storage (vector) · same retrieval (similarity) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also open source · also self-hostable · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ? ★ 3.2k released 2026-04-08 MIT Compare with Letta / MemGPT →

  6. 6 Botpress LLMz Framework-embedded memory

    same storage (vector) · same retrieval (similarity, injection) · same conflict handling (llm-arbitrate) · same memory unit (episode) · also open source · also self-hostable · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 15k MIT (SDK/integration packages in… Compare with Letta / MemGPT →

  7. 7 O-Mem Dedicated memory layers

    also in Dedicated memory layers · same storage (vector) · same retrieval (similarity) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also open source · also self-hostable

    OSS ✓ Self-host ✓ Managed ✓ MCP ? ★ 79 Apache-2.0 Compare with Letta / MemGPT →

  8. 8 LangMem (LangChain) Framework-embedded memory

    same storage (kv, vector) · same retrieval (similarity, injection) · same conflict handling (llm-arbitrate) · same memory unit (episode, fact) · also open source · also self-hostable · also offered managed

    OSS ✓ Self-host ✓ Managed ✓ MCP ✓ ★ 1.6k released 2025-10-27 MIT Compare with Letta / MemGPT →

Ranked by shared memory design (storage, retrieval, conflict handling, memory unit), same section and shared must-haves (open source, self-hosting, MCP), with adoption as a tie-breaker. ✓ confirmed · — no · ? not in the catalog yet. Want picks for your own requirements? Use the chooser. Checking a whole project? Check your stack.