OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

Deepmem

Deepmem is a Python library and self-hostable service that gives AI agents and LLM applications searchable long-term memory. Its FastAPI HTTP API uses Qdrant and supports LLM fact extraction, hybrid retrieval, and a built-in MCP server; a managed cloud service is also documented.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Deepmem — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Deepmem when an agent or LLM application needs persistent conversation memory and you want a self-hostable service or a managed cloud option. It is also relevant when adapting a Mem0 client using the documented MemoryClient interface.

Who it is for

  • Developers adding persistent, searchable memory to AI agents or LLM applications.
  • Teams migrating a Mem0 client to a Mem0-compatible interface.
  • Operators who want to self-host an AI memory service or use its managed cloud option.

Documented uses

  • Store conversation content as memory for later retrieval.
  • Search stored memory using vector and BM25 retrieval, with entity boosts and time decay.
  • Use LLM fact extraction to turn conversations into memory.
  • Connect applications through the HTTP API, Python client, or built-in MCP server.
Before you choose

Requirements & limits

The repository is MIT-licensed and documents self-hosting through Docker or from source. The managed cloud uses an API key. Self-hosted raw storage and retrieval are documented as working without network access; LLM fact extraction requires an LLM key.

  • Self-hosting requires configuring an LLM key for fact extraction; the README also documents embedding-provider and Qdrant configuration options.
  • The client interface is documented as Mem0-compatible, but graph relations are not replicated.
  • Deepmem’s reset operation is scoped per user ID, unlike the account-wide reset described for Mem0.
  • The self-hosted setup exposes HTTP on port 8000 and MCP on port 8001.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run Deepmem with Docker

Start the self-hosted HTTP and MCP services with the documented Docker Compose setup.

Before you start

  • The Deepmem repository and Docker Compose
  • An LLM key added to the environment configuration

Steps

  1. Copy .env.example to .env.

  2. Add an LLM key to .env.

  3. Run the command below.

    Documented command
    docker compose up --build

  4. Check the HTTP service at http://localhost:8000/health.

Expected outcome

The documented setup starts Deepmem’s HTTP service on port 8000, MCP service on port 8001, and a Qdrant sidecar.

Guide limitations

  • The guide documents startup, not a verified deployment or successful runtime test.
  • LLM fact extraction requires an LLM key; raw storage and retrieval are documented as working without network access.
Documentation 1 ↗
Source-based practical guide

Replace a Mem0 client import

Adapt a Mem0-based client to use Deepmem’s documented `MemoryClient` interface.

Before you start

  • An application using mem0.MemoryClient
  • The deepmem-client package

Steps

  1. Install the package with the command below.

    Documented command
    pip install deepmem-client

  2. Replace from mem0 import MemoryClient with from deepmem import MemoryClient.

  3. Use the documented Deepmem API key when configuring the managed cloud client.

Expected outcome

The README demonstrates retaining calls such as `add` and `search` after changing the import.

Guide limitations

  • The compatibility claim covers the documented method signatures, not execution testing.
  • Deepmem does not replicate Mem0 graph relations, and reset is per user_id rather than account-wide.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
deepmemteam/deepmem · Official homepage title · retained in its source language
How we review sourcesSuggest a correction

Related options

Mem0 — Deepmem documents a Mem0-compatible MemoryClient interface and a migration path that replaces the import. It notes that graph relations and account-wide reset behavior are not repl

Explore project →