FalkorDB
Ultra-fast graph database purpose-built for GraphRAG and agent memory.
About
FalkorDB is an open-source graph database built on GraphBLAS sparse-matrix algebra and running as a Redis module, making it the direct continuation of RedisGraph after Redis Labs discontinued it in early 2025. Founded in Tel Aviv in 2023, it raised a $3M seed from Angular Ventures and positions itself as the knowledge-graph layer for LLM applications.
Its AI play is explicit: a first-party GraphRAG SDK that ranked first on GraphRAG-Bench across all four question types in its 1.0 release (April 2026), an MCP server for LLM integration, mem0 integration for persistent agent memory, and a September 2026 v6.0 release rebuilt on a new Rust engine with 39% fewer CPU instructions. Query latency claims center on sub-10ms multi-hop traversals.
Limitations to note: it is licensed SSPL-1.0, which is not OSI-approved and restricts offering it as a managed service; the ecosystem is far smaller than Neo4j's; and benchmark wins are vendor-run and should be reproduced on your own graph shapes.
What it can do
- GraphBLAS-powered low-latency graph queries
- GraphRAG SDK 1.0 for knowledge-graph pipelines
- MCP server and LangChain/LangGraph integrations
- Built-in vector indexing and full-text search
Setup
Run via Docker as a Redis module, or use FalkorDB Cloud with a free tier for MVPs.
Prerequisites
Familiarity with Cypher/OpenCypher for querying; Docker for self-hosting.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 83; Trending score 76/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.