Embeddable graph database built in pure Rust. Dual data models (LPG + RDF), six query languages, HNSW vector search, MVCC transactions. Fastest graph database in our graph-bench suite (which includes workloads inspired by the LDBC Social Network Benchmark), in both embedded and server configurations, with a low memory footprint. Now in beta.
Try it live at grafeo.ai
- 6 query languages: GQL (ISO), Cypher, Gremlin, GraphQL, SPARQL, SQL/PGQ. Use what you know or want to learn.
- LPG + RDF in one engine: Property graphs and triples with optimized storage for each.
- Vector search: HNSW indexes with scalar, binary, and product quantization. SIMD-accelerated.
- Embeddable: Link as a library (Rust, Python, Node.js, Go, C#, Dart, WASM) or run as a server. No external processes.
- Fast and lean: Fastest graph database in our graph-bench suite (includes LDBC-inspired workloads) in both embedded and server modes, with a lower memory footprint than other in-memory databases. Push-based vectorized execution, morsel-driven parallelism, columnar storage with zone maps. Also benchmarked with ann-benchmarks.
import grafeo
db = grafeo.GrafeoDB()
# Build a graph
db.execute("INSERT (:Person {name: 'Alice', age: 30})-[:KNOWS {since: 2020}]->(:Person {name: 'Bob', age: 25})")
db.execute("INSERT (:Person {name: 'Bob'})-[:KNOWS]->(:Person {name: 'Carol', age: 28})")
# Traverse it
for row in db.execute("""
MATCH (p:Person)-[:KNOWS]->(friend)-[:KNOWS]->(fof)
WHERE fof <> p
RETURN p.name, fof.name AS friend_of_friend
"""):
print(row)
| Project |
Description |
Install |
| grafeo |
Embeddable graph database engine |
uv add grafeo / cargo add grafeo / npm install @grafeo-db/js / dotnet add package Grafeo / dart pub add grafeo |
| grafeo-server |
HTTP server & web UI: REST API, transactions, ~40MB Docker image |
docker pull grafeo/grafeo-server |
| grafeo-web |
Browser-based Grafeo via WebAssembly with IndexedDB persistence |
npm install @grafeo-db/web |
| Project |
Description |
Install |
| anywidget-graph |
Interactive graph visualization for notebooks (Marimo, Jupyter, VS Code, Colab) |
uv add anywidget-graph |
| anywidget-vector |
3D vector/embedding visualization with 6D encoding |
uv add anywidget-vector |
| Project |
Description |
Install |
| grafeo-memory |
AI memory layer. Extract, reconcile, and search memories from conversations. No Docker, no Neo4j, one .db file + one LLM |
uv add grafeo-memory |
| grafeo-mcp |
Use Grafeo as an MCP server, expose to AI agents over stdio or HTTP |
uv add grafeo-mcp |
| grafeo-langchain |
LangChain GraphStore & GraphVectorStore for knowledge-graph RAG |
uv add grafeo-langchain |
| grafeo-llamaindex |
LlamaIndex PropertyGraphStore with structured + vector queries |
uv add grafeo-llamaindex |
Wire Protocols (for Grafeo Server)
| Project |
Description |
Install |
| gwp |
GQL Wire Protocol: gRPC wire protocol for GQL (ISO/IEC 39075) with client bindings in 5 languages |
cargo add gwp |
| boltr |
Bolt Wire Protocol: pure Rust Bolt v5.x implementation for Neo4j driver compatibility |
cargo add boltr |
| Project |
Description |
| playground |
Interactive browser playground: query in 6 languages, visualize graphs, explore schemas, all client-side via WebAssembly |
| Project |
Description |
| graph-bench |
Benchmark suite: 65 benchmarks (LDBC-inspired SNB, graph analytics, ACID, vector, hybrid) across 10 graph database engines |
| ann-benchmarks |
Fork of ann-benchmarks extended with a Grafeo HNSW adapter for vector search benchmarking |