TerminusDB is the open-source knowledge graph database with git-like version control for graphs and vectors. Branch, merge, diff and hybrid search any commit.
3,400+ GitHub stars (GitHub) · 850,000+ Docker pulls (Docker Hub) · latest release 12.0.9 · Release 12.1 coming soon! — as of November 2026.
What makes it different
- Git for your graph. Branch, diff and merge your data, then review the change before it lands. How knowledge graph version control works →
- Time travel and audit. Every write is an immutable commit, so you can query any point in history and see who changed what. Git-like version control for graph data →
- Hybrid and vector search with VectorLink. Search your documents by meaning and by structure, as of any commit. How to use VectorLink →
- Documents and graph in one store. JSON documents and RDF triples in one database, queried with Datalog (WOQL) or GraphQL. About the documents API →
Use cases
AI agents and GraphRAG
Give agents a memory they can trust: vector and hybrid search over versioned documents, as of any commit.
Regulated reference data
Every write is an immutable commit with author and timestamp, so audits answer who changed what, and when.
Product and configuration data
A schema-enforced document graph keeps catalogues and configuration consistent on every write.
Research and history
TerminusDB began as the store behind the Seshat Global History Databank, built for contested, evolving records.
Three ways to run it
- Self-host, free. Docker on your own machine or server. Start in 5 minutes →
- Azure Marketplace. Pay as you go, inside your own Azure tenant. Not live yet — join the waitlist.
- Managed or design partner. We run it with you. Talk to us.
Quickstart in 5 minutes
docker run --pull always -d -p 127.0.0.1:6363:6363 \
-v terminusdb_storage:/app/terminusdb/storage \
--name terminusdb terminusdb/terminusdb-server:v12Then clone a ready-made dataset and diff your first branch: First 10 minutes →
Where it comes from
TerminusDB started as the data store for the Seshat Global History Databank, where historians needed to track who said what about the past, and change their minds without losing the old version.
Community
Ask questions and share what you build on Discord or GitHub.
TerminusDB is maintained and offered by Gettier since 2025. We provide the simplest way for subject matter experts to build AI-ready, defensible data foundations. Gettier is a proof layer powered by TerminusDB. We offer Enterprises what they need to run TerminusDB data foundations in production, with or without the Gettier proof layer.
Start in 5 minutes · Talk to us
Frequently asked questions
What is TerminusDB?
TerminusDB is the open-source knowledge graph database with git-like version control for graphs and vectors. Branch, merge, diff and hybrid search any commit. It runs in a single Docker container.
How is TerminusDB different from Neo4j?
Neo4j is a property graph database. TerminusDB stores JSON documents as an RDF graph and keeps every change as a commit, so you can branch, diff, merge and query any point in history.
Is TerminusDB open source?
Yes. TerminusDB is licensed under Apache 2.0 and free to self-host.
What does TerminusDB cost?
Self-hosting is free. A pay-as-you-go Azure Marketplace offer and managed options are coming; talk to us for details.
What is VectorLink?
VectorLink is the vector and hybrid search built into TerminusDB. It lets you search documents by meaning and by structure, as of any commit.