databases.json

{
  "_meta": {
    "version": "3.0.0",
    "last_updated": "2026-04-10",
    "title": "Graph Database Resource Catalog",
    "scope": "Primary-source reference collection of RDF stores, label property graph databases, multi-model engines, and adjacent graph systems with explicit vector-strategy classification.",
    "methodology": "Each entry is anchored to current official product pages, documentation, or source repositories. The catalog distinguishes native vector search from connector-based, sibling-service, and paired-extension patterns. Public-doc ambiguity is labeled explicitly instead of inferred as a capability.",
    "models": {
      "rdf": "RDF / SPARQL-native system",
      "property": "Label property graph database or managed property-graph service",
      "multi": "Multi-model system that includes graph alongside other models",
      "specialized": "Graph-adjacent or specialized knowledge engine that does not map cleanly to RDF or LPG"
    },
    "statuses": {
      "production": "Actively shipped and suitable for serious evaluation or deployment",
      "emerging": "Active but still early, preview, or insufficiently battle-tested",
      "legacy": "Historical, archived, end-of-life, or not recommended for new deployments"
    },
    "vector_strategies": {
      "native": "Vector indexing and query capability are documented in the same engine",
      "integrated": "Vector capability depends on an integrated search subsystem or connector",
      "paired": "The documented pattern is to pair the graph engine with an extension, toolkit, or sibling component in the same stack",
      "sibling": "A separate managed service or API in the same vendor platform provides vector search",
      "none": "No documented vector capability in the graph engine",
      "unclear": "AI or semantic-search claims exist, but current public docs do not expose a clear general-purpose vector implementation"
    },
    "license_types": {
      "permissive": "Permissive open-source license",
      "copyleft": "Copyleft open-source license",
      "open-core": "Open-source core with commercial edition for key capabilities",
      "source-available": "Source available or community license with usage restrictions",
      "commercial": "Commercial or managed-service offering"
    }
  },
  "databases": [
    {
      "id": "qlever",
      "name": "QLever",
      "model": "rdf",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://github.com/ad-freiburg/qlever",
      "query_interfaces": ["SPARQL 1.1"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; pair with an external vector engine when needed.",
      "ha": "No native cluster",
      "deployment": "Self-hosted, single-node",
      "best_fit": "High-performance open-source SPARQL on large datasets",
      "profile": "databases/qlever/README.md",
      "source_urls": [
        "https://github.com/ad-freiburg/qlever",
        "https://github.com/ad-freiburg/qlever/wiki/QLever-performance-evaluation-and-comparison-to-other-SPARQL-engines"
      ],
      "notes": "Strong benchmark story and production endpoints; vector search is out of scope for the core engine."
    },
    {
      "id": "apache-jena",
      "name": "Apache Jena / Fuseki",
      "model": "rdf",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://jena.apache.org",
      "query_interfaces": ["SPARQL 1.1"],
      "vector_strategy": "none",
      "vector_summary": "No native ANN vector feature in Jena or Fuseki; use an external vector service.",
      "ha": "Depends on external replication pattern",
      "deployment": "Self-hosted",
      "best_fit": "Standards-heavy RDF, reasoning, and Java ecosystem integration",
      "profile": "databases/apache-jena/README.md",
      "source_urls": [
        "https://jena.apache.org/documentation/fuseki2/",
        "https://jena.apache.org/documentation/tdb2/"
      ],
      "notes": "Broadest semantic-web ecosystem in the catalog."
    },
    {
      "id": "virtuoso-ose",
      "name": "OpenLink Virtuoso OSE",
      "model": "multi",
      "status": "production",
      "license": "GPL v2",
      "license_type": "copyleft",
      "website": "https://github.com/openlink/virtuoso-opensource",
      "query_interfaces": ["SPARQL 1.1", "SQL"],
      "vector_strategy": "none",
      "vector_summary": "Open-source Virtuoso is not documented as a modern native ANN vector engine.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Battle-tested RDF plus SQL in one engine",
      "profile": "databases/virtuoso-ose/README.md",
      "source_urls": [
        "https://github.com/openlink/virtuoso-opensource",
        "https://www.w3.org/wiki/LargeTripleStores"
      ],
      "notes": "Still one of the safest choices when scale and SPARQL maturity matter more than vector support."
    },
    {
      "id": "oxigraph",
      "name": "Oxigraph",
      "model": "rdf",
      "status": "emerging",
      "license": "MIT / Apache 2.0",
      "license_type": "permissive",
      "website": "https://github.com/oxigraph/oxigraph",
      "query_interfaces": ["SPARQL 1.1", "SPARQL 1.2 drafts"],
      "vector_strategy": "none",
      "vector_summary": "No vector index is documented in the core project.",
      "ha": "No",
      "deployment": "Embedded library, self-hosted server, WASM",
      "best_fit": "Embeddable Rust RDF workloads",
      "profile": "databases/oxigraph/README.md",
      "source_urls": [
        "https://github.com/oxigraph/oxigraph",
        "https://pyoxigraph.readthedocs.io/"
      ],
      "notes": "Excellent embedding story, but still better treated as emerging than fully mature infrastructure."
    },
    {
      "id": "sutradb",
      "name": "SutraDB",
      "model": "rdf",
      "status": "emerging",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://sutradb.org",
      "query_interfaces": ["SPARQL 1.1 superset"],
      "vector_strategy": "native",
      "vector_summary": "Embeddings are stored in RDF and indexed natively with HNSW-style vector search.",
      "ha": "No",
      "deployment": "Embedded / self-hosted",
      "best_fit": "Early RDF plus vector experiments in one query model",
      "profile": "databases/sutradb/README.md",
      "source_urls": [
        "https://sutradb.org"
      ],
      "notes": "Interesting architecture, but still preview-stage."
    },
    {
      "id": "oxirs",
      "name": "OxiRS",
      "model": "rdf",
      "status": "emerging",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://github.com/cool-japan/oxirs",
      "query_interfaces": ["SPARQL", "GraphQL"],
      "vector_strategy": "native",
      "vector_summary": "The project documents vector support through its own crates, but the public evidence is still mostly self-reported.",
      "ha": "Claimed",
      "deployment": "Self-hosted",
      "best_fit": "Experimental Rust semantic stack",
      "profile": "databases/oxirs/README.md",
      "source_urls": [
        "https://github.com/cool-japan/oxirs"
      ],
      "notes": "Treat vendor README claims carefully until more independent evidence exists."
    },
    {
      "id": "grafeo",
      "name": "Grafeo",
      "model": "multi",
      "status": "emerging",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://grafeo.dev",
      "query_interfaces": ["GQL", "Cypher", "Gremlin", "SPARQL"],
      "vector_strategy": "native",
      "vector_summary": "The product documents native vector search, but the public SPARQL depth and production evidence remain limited.",
      "ha": "Not clearly documented",
      "deployment": "Embedded / self-hosted",
      "best_fit": "Polyglot graph R&D and API experimentation",
      "profile": "databases/grafeo/README.md",
      "source_urls": [
        "https://grafeo.dev"
      ],
      "notes": "Interesting multi-model direction, still early for production benchmarking."
    },
    {
      "id": "stardog",
      "name": "Stardog",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.stardog.com",
      "query_interfaces": ["SPARQL 1.1", "GraphQL"],
      "vector_strategy": "native",
      "vector_summary": "Vector search is documented inside the same Stardog stack and query surface.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed options via partners",
      "best_fit": "Enterprise RDF, reasoning, and GraphRAG in one platform",
      "profile": "databases/stardog/README.md",
      "source_urls": [
        "https://docs.stardog.com/query-stardog/semantic-search",
        "https://docs.stardog.com/query-stardog/full-text-search"
      ],
      "notes": "One of the clearest commercial RDF plus vector offerings."
    },
    {
      "id": "allegrograph",
      "name": "AllegroGraph",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://allegrograph.com",
      "query_interfaces": ["SPARQL 1.1", "Prolog"],
      "vector_strategy": "native",
      "vector_summary": "VectorStore is documented as a native AllegroGraph feature.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Commercial RDF, reasoning, and graph-plus-vector use cases",
      "profile": "databases/allegrograph/README.md",
      "source_urls": [
        "https://franz.com/agraph/support/documentation/vector-dbs-and-rag.html",
        "https://allegrograph.com"
      ],
      "notes": "Long-running commercial RDF store with explicit graph-plus-vector positioning."
    },
    {
      "id": "graphdb",
      "name": "GraphDB",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://graphdb.ontotext.com",
      "query_interfaces": ["SPARQL 1.1"],
      "vector_strategy": "integrated",
      "vector_summary": "Vector search is delivered through the Elasticsearch or OpenSearch connector layer, not as an in-engine ANN index.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed by partners",
      "best_fit": "Reasoning-heavy RDF where connector-based vector search is acceptable",
      "profile": "databases/graphdb/README.md",
      "source_urls": [
        "https://graphdb.ontotext.com/documentation/11.3/embedding-model-configuration.html",
        "https://graphdb.ontotext.com/documentation/11.3/"
      ],
      "notes": "Important distinction: integrated vector pattern, not native vector core."
    },
    {
      "id": "amazon-neptune",
      "name": "Amazon Neptune",
      "model": "multi",
      "status": "production",
      "license": "Commercial managed service",
      "license_type": "commercial",
      "website": "https://aws.amazon.com/neptune/",
      "query_interfaces": ["SPARQL 1.1", "Gremlin", "openCypher"],
      "vector_strategy": "sibling",
      "vector_summary": "Vector search is exposed through Neptune Analytics rather than the core Neptune database engine.",
      "ha": "Managed",
      "deployment": "AWS managed service",
      "best_fit": "Managed RDF plus property graph on AWS with adjacent analytics",
      "profile": "databases/amazon-neptune/README.md",
      "source_urls": [
        "https://docs.aws.amazon.com/neptune-analytics/",
        "https://docs.aws.amazon.com/neptune/latest/userguide/sparql.html"
      ],
      "notes": "Useful when AWS-managed operations matter more than single-engine purity."
    },
    {
      "id": "rdfox",
      "name": "RDFox",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.oxfordsemantic.tech/rdfox",
      "query_interfaces": ["SPARQL 1.1", "Datalog"],
      "vector_strategy": "unclear",
      "vector_summary": "Current public docs clearly document RDF, SPARQL, and reasoning. A general-purpose vector feature is not clearly exposed in the public technical docs we could verify.",
      "ha": "Yes",
      "deployment": "Self-hosted, embeddable",
      "best_fit": "High-performance in-memory RDF and rules-based reasoning",
      "profile": "databases/rdfox/README.md",
      "source_urls": [
        "https://docs.oxfordsemantic.tech/",
        "https://www.oxfordsemantic.tech/rdfox"
      ],
      "notes": "Do not classify RDFox as a confirmed vector database without version-specific vendor evidence."
    },
    {
      "id": "marklogic",
      "name": "MarkLogic",
      "model": "multi",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.progress.com/marklogic",
      "query_interfaces": ["SPARQL 1.1", "SQL", "XQuery", "JavaScript"],
      "vector_strategy": "native",
      "vector_summary": "MarkLogic documents built-in vector search alongside graph, search, and document features.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Enterprise multi-model with graph, search, and vector in one platform",
      "profile": "databases/marklogic/README.md",
      "source_urls": [
        "https://www.progress.com/marklogic/server/features/vector-search",
        "https://docs.progress.com/bundle/marklogic-server-develop-12/page/topics/semantics.html"
      ],
      "notes": "Strong option when one enterprise platform must cover search, documents, graph, and vector."
    },
    {
      "id": "surrealdb",
      "name": "SurrealDB",
      "model": "multi",
      "status": "production",
      "license": "BSL 1.1",
      "license_type": "source-available",
      "website": "https://surrealdb.com/docs/surrealdb",
      "query_interfaces": ["SurrealQL", "GraphQL"],
      "vector_strategy": "native",
      "vector_summary": "SurrealDB documents brute-force and HNSW vector search inside SurrealQL.",
      "ha": "Yes",
      "deployment": "Embedded, self-hosted, managed cloud",
      "best_fit": "Unified application database with graph and vector in one query language",
      "profile": "databases/surrealdb/README.md",
      "source_urls": [
        "https://surrealdb.com/docs/surrealdb/reference-guide/vector-search",
        "https://surrealdb.com/docs/surrealdb"
      ],
      "notes": "One of the cleaner multi-model vector stories in current public docs."
    },
    {
      "id": "terminusdb",
      "name": "TerminusDB + VectorLink",
      "model": "multi",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://terminusdb.org",
      "query_interfaces": ["WOQL", "GraphQL"],
      "vector_strategy": "paired",
      "vector_summary": "VectorLink is a separate service paired with TerminusDB rather than a native graph-engine feature.",
      "ha": "Limited",
      "deployment": "Self-hosted",
      "best_fit": "Versioned knowledge-graph workflows with a sidecar vector service",
      "profile": "databases/terminusdb/README.md",
      "source_urls": [
        "https://terminusdb.org",
        "https://github.com/terminusdb-labs/vectorlink"
      ],
      "notes": "Strong for document and versioning workflows, but not a single-engine RDF plus vector story."
    },
    {
      "id": "blazegraph",
      "name": "Blazegraph",
      "model": "rdf",
      "status": "legacy",
      "license": "GPL v2",
      "license_type": "copyleft",
      "website": "https://github.com/blazegraph/database",
      "query_interfaces": ["SPARQL 1.1"],
      "vector_strategy": "none",
      "vector_summary": "No native vector search is documented.",
      "ha": "Historical",
      "deployment": "Self-hosted",
      "best_fit": "Legacy RDF migration planning",
      "profile": "databases/blazegraph/README.md",
      "source_urls": [
        "https://github.com/blazegraph/database"
      ],
      "notes": "Keep only as a legacy reference, not a recommendation for new deployments."
    },
    {
      "id": "neo4j",
      "name": "Neo4j",
      "model": "property",
      "status": "production",
      "license": "GPL v3 Community / commercial Enterprise",
      "license_type": "open-core",
      "website": "https://neo4j.com",
      "query_interfaces": ["Cypher", "GQL"],
      "vector_strategy": "native",
      "vector_summary": "Lucene-backed HNSW vector indexes are documented. The first-class VECTOR type requires server 2025.10+ Enterprise.",
      "ha": "Enterprise / Aura",
      "deployment": "Self-hosted, Aura managed",
      "best_fit": "Cypher-centric OLTP graph workloads and enterprise GraphRAG",
      "profile": "databases/neo4j/README.md",
      "source_urls": [
        "https://neo4j.com/docs/cypher-manual/current/indexes/semantic-indexes/vector-indexes/",
        "https://neo4j.com/docs/go-manual/current/data-types/"
      ],
      "notes": "Largest ecosystem in the LPG segment."
    },
    {
      "id": "memgraph",
      "name": "Memgraph",
      "model": "property",
      "status": "production",
      "license": "BSL 1.1",
      "license_type": "source-available",
      "website": "https://memgraph.com",
      "query_interfaces": ["Cypher", "GQL"],
      "vector_strategy": "native",
      "vector_summary": "Official releases document HNSW-style vector indexes for both nodes and relationships.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Real-time graph workloads with integrated vector search",
      "profile": "databases/memgraph/README.md",
      "source_urls": [
        "https://memgraph.com/blog/memgraph-3-4-release-announcement",
        "https://memgraph.com/docs"
      ],
      "notes": "Relationship vectors are a differentiator for recommendation and interaction-heavy workloads."
    },
    {
      "id": "falkordb",
      "name": "FalkorDB",
      "model": "property",
      "status": "production",
      "license": "SSPL v1",
      "license_type": "source-available",
      "website": "https://www.falkordb.com",
      "query_interfaces": ["Cypher", "Bolt procedures"],
      "vector_strategy": "native",
      "vector_summary": "Vector indexes are documented for both nodes and relationships, with HNSW parameters exposed.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed options",
      "best_fit": "Fast graph plus vector search with RedisGraph lineage",
      "profile": "databases/falkordb/README.md",
      "source_urls": [
        "https://docs.falkordb.com/cypher/indexing/vector-index",
        "https://www.falkordb.com"
      ],
      "notes": "Active successor path for teams leaving RedisGraph."
    },
    {
      "id": "arangodb",
      "name": "ArangoDB",
      "model": "multi",
      "status": "production",
      "license": "ArangoDB Community License / commercial Enterprise",
      "license_type": "source-available",
      "website": "https://arangodb.com",
      "query_interfaces": ["AQL"],
      "vector_strategy": "native",
      "vector_summary": "ArangoDB 3.12 documents native vector indexes and combines them with graph traversals in AQL.",
      "ha": "Yes",
      "deployment": "Self-hosted, ArangoGraph managed cloud",
      "best_fit": "Teams that want graphs, search, documents, and vectors in one query model",
      "profile": "databases/arangodb/README.md",
      "source_urls": [
        "https://docs.arangodb.com/3.12/develop/http-api/indexes/vector/",
        "https://docs.arangodb.com/3.12/get-started/start-using-aql/graphs/",
        "https://docs.arangodb.com/3.13/about-arangodb/features/"
      ],
      "notes": "Current community licensing is not permissive open source; read the community license carefully."
    },
    {
      "id": "arcadedb",
      "name": "ArcadeDB",
      "model": "multi",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://arcadedb.com",
      "query_interfaces": ["SQL", "openCypher", "Gremlin", "GraphQL", "MongoDB protocol"],
      "vector_strategy": "native",
      "vector_summary": "ArcadeDB publicly documents vectors as a first-class built-in model inside the same engine.",
      "ha": "Yes",
      "deployment": "Embedded, self-hosted, Kubernetes",
      "best_fit": "Polyglot multi-model workloads that need a genuinely open license",
      "profile": "databases/arcadedb/README.md",
      "source_urls": [
        "https://arcadedb.com/",
        "https://arcadedb.com/knowledge-graphs.html"
      ],
      "notes": "Strong open-source multi-model option if you accept a younger ecosystem than Neo4j or ArangoDB."
    },
    {
      "id": "nebulagraph",
      "name": "NebulaGraph",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0 Community / commercial Enterprise",
      "license_type": "open-core",
      "website": "https://www.nebula-graph.io",
      "query_interfaces": ["nGQL", "openCypher"],
      "vector_strategy": "native",
      "vector_summary": "NebulaGraph documents native vector search in Enterprise and newer cloud offerings, not in the Apache 2.0 community core.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed cloud",
      "best_fit": "Distributed LPG at large scale, especially in enterprise deployments",
      "profile": "databases/nebulagraph/README.md",
      "source_urls": [
        "https://www.nebula-graph.io/posts/NebulaGraph_v5.1_Embeds_Vector_Search",
        "https://www.nebula-graph.io/posts/NebulaGraph_Cloud_Latest_Version"
      ],
      "notes": "Vector classification here reflects enterprise and cloud docs, not community-only deployments."
    },
    {
      "id": "hugegraph",
      "name": "Apache HugeGraph",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://hugegraph.apache.org",
      "query_interfaces": ["Gremlin", "Cypher", "REST"],
      "vector_strategy": "paired",
      "vector_summary": "GraphRAG and embedding workflows are documented through HugeGraph-AI and HugeGraph-LLM rather than the OLTP core database.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Large-scale Gremlin workloads with adjacent GraphRAG tooling",
      "profile": "databases/hugegraph/README.md",
      "source_urls": [
        "https://hugegraph.apache.org/docs/",
        "https://hugegraph.apache.org/docs/quickstart/hugegraph-ai/",
        "https://hugegraph.apache.org/docs/quickstart/hugegraph-ai/hugegraph-llm/"
      ],
      "notes": "Do not mistake the AI toolkit for a native vector index in the core server."
    },
    {
      "id": "apache-age",
      "name": "Apache AGE",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://age.apache.org/overview/",
      "query_interfaces": ["openCypher", "SQL"],
      "vector_strategy": "paired",
      "vector_summary": "The practical vector pattern is AGE plus PostgreSQL's pgvector extension in the same database stack.",
      "ha": "PostgreSQL dependent",
      "deployment": "Self-hosted, managed PostgreSQL variants",
      "best_fit": "PostgreSQL shops adding graph queries without leaving Postgres",
      "profile": "databases/apache-age/README.md",
      "source_urls": [
        "https://age.apache.org/overview/",
        "https://github.com/pgvector/pgvector"
      ],
      "notes": "This is an architectural pairing, not a built-in AGE feature."
    },
    {
      "id": "tigergraph",
      "name": "TigerGraph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.tigergraph.com",
      "query_interfaces": ["GSQL"],
      "vector_strategy": "native",
      "vector_summary": "TigerGraph 4.2 release notes document hybrid vector search with native vector attributes and ANN indexing.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed cloud",
      "best_fit": "Large-scale operational property graphs with integrated vector search",
      "profile": "databases/tigergraph/README.md",
      "source_urls": [
        "https://docs.tigergraph.com/tigergraph-server/4.2/release-notes/"
      ],
      "notes": "Production-ready, but firmly commercial."
    },
    {
      "id": "dgraph",
      "name": "Dgraph",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://dgraph.io",
      "query_interfaces": ["DQL", "GraphQL"],
      "vector_strategy": "native",
      "vector_summary": "Dgraph documents float32vector plus HNSW indexes in DQL.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed by partners",
      "best_fit": "Distributed graph plus vector with GraphQL-friendly workflows",
      "profile": "databases/dgraph/README.md",
      "source_urls": [
        "https://docs.dgraph.io/learn/howto/similarity-search/"
      ],
      "notes": "Clear first-class vector support in current official docs."
    },
    {
      "id": "janusgraph",
      "name": "JanusGraph",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://janusgraph.org",
      "query_interfaces": ["Gremlin"],
      "vector_strategy": "none",
      "vector_summary": "Current public JanusGraph docs emphasize graph plus search backends, not a native vector index.",
      "ha": "Backend dependent",
      "deployment": "Self-hosted",
      "best_fit": "Distributed Gremlin over Cassandra, HBase, Bigtable, or similar backends",
      "profile": "databases/janusgraph/README.md",
      "source_urls": [
        "https://janusgraph.org/",
        "https://docs.janusgraph.org/"
      ],
      "notes": "Still relevant when you want a graph layer on top of existing backend infrastructure."
    },
    {
      "id": "orientdb",
      "name": "OrientDB",
      "model": "multi",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://orientdb.dev",
      "query_interfaces": ["SQL", "Gremlin"],
      "vector_strategy": "none",
      "vector_summary": "Current public OrientDB docs do not document a native vector engine.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "General-purpose multi-model graph and document workloads",
      "profile": "databases/orientdb/README.md",
      "source_urls": [
        "https://orientdb.dev",
        "https://github.com/orientechnologies/orientdb"
      ],
      "notes": "Still active, but less differentiated than newer multi-model graph systems."
    },
    {
      "id": "typedb",
      "name": "TypeDB",
      "model": "specialized",
      "status": "production",
      "license": "Polyform / MPL by component",
      "license_type": "source-available",
      "website": "https://typedb.com",
      "query_interfaces": ["TypeQL"],
      "vector_strategy": "none",
      "vector_summary": "Current public TypeDB docs emphasize reasoning and schema, not vector indexing.",
      "ha": "Yes",
      "deployment": "Self-hosted, TypeDB Cloud",
      "best_fit": "Strongly typed reasoning-heavy knowledge systems",
      "profile": "databases/typedb/README.md",
      "source_urls": [
        "https://typedb.com/docs/home/get-started/overview",
        "https://github.com/typedb/typedb"
      ],
      "notes": "Included as a specialized reference point, not as a standard LPG."
    },
    {
      "id": "cosmos-gremlin",
      "name": "Azure Cosmos DB for Gremlin",
      "model": "property",
      "status": "production",
      "license": "Commercial managed service",
      "license_type": "commercial",
      "website": "https://learn.microsoft.com/azure/cosmos-db/gremlin/",
      "query_interfaces": ["Gremlin"],
      "vector_strategy": "sibling",
      "vector_summary": "Cosmos vector indexing is documented for NoSQL, not for the Gremlin API itself.",
      "ha": "Managed",
      "deployment": "Azure managed service",
      "best_fit": "Managed Gremlin on Azure when vector search can live in a sibling Cosmos API",
      "profile": "databases/cosmos-gremlin/README.md",
      "source_urls": [
        "https://learn.microsoft.com/en-us/azure/cosmos-db/gremlin/overview",
        "https://learn.microsoft.com/en-us/azure/cosmos-db/vector-search"
      ],
      "notes": "Important API boundary: Gremlin and vector search are not the same Cosmos API surface."
    },
    {
      "id": "puppygraph",
      "name": "PuppyGraph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.puppygraph.com",
      "query_interfaces": ["Gremlin", "openCypher"],
      "vector_strategy": "paired",
      "vector_summary": "PuppyGraph virtualizes graph queries over external stores; vector behavior is delegated to those underlying systems.",
      "ha": "Yes",
      "deployment": "Self-hosted, cloud marketplace images",
      "best_fit": "Zero-ETL graph analytics over warehouse or lakehouse data",
      "profile": "databases/puppygraph/README.md",
      "source_urls": [
        "https://docs.puppygraph.com/",
        "https://docs.puppygraph.com/querying/",
        "https://docs.puppygraph.com/reference/cypher-query-language/"
      ],
      "notes": "Architecturally important because it answers whether you need a separate graph store at all."
    },
    {
      "id": "kuzu",
      "name": "Kuzu",
      "model": "property",
      "status": "legacy",
      "license": "MIT",
      "license_type": "permissive",
      "website": "https://github.com/kuzudb/kuzu",
      "query_interfaces": ["Cypher"],
      "vector_strategy": "native",
      "vector_summary": "Kuzu shipped native vector and full-text extensions before the repository was archived.",
      "ha": "No",
      "deployment": "Embedded",
      "best_fit": "Historical reference for embedded analytical property graphs",
      "profile": "databases/kuzu/README.md",
      "source_urls": [
        "https://github.com/kuzudb/kuzu"
      ],
      "notes": "Useful only as a legacy benchmark reference and migration case."
    },
    {
      "id": "ultipa",
      "name": "Ultipa",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.ultipa.com",
      "query_interfaces": ["GQL", "Ultipa APIs"],
      "vector_strategy": "native",
      "vector_summary": "Ultipa documents vector indexes and vector query procedures in the graph engine.",
      "ha": "Yes",
      "deployment": "Embedded, self-hosted, managed cloud",
      "best_fit": "High-scale property graph workloads with strong AI positioning",
      "profile": "databases/ultipa/README.md",
      "source_urls": [
        "https://www.ultipa.com/docs/gql/vector-index",
        "https://www.ultipa.com/product/ultipa-graph",
        "https://www.ultipa.com/article/technical/ultipa-now-supports-vector-search"
      ],
      "notes": "Active and credible, but mostly vendor-supplied public evidence."
    },
    {
      "id": "aerospike",
      "name": "Aerospike Graph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://aerospike.com/docs/graph/quick-start",
      "query_interfaces": ["Gremlin"],
      "vector_strategy": "paired",
      "vector_summary": "Aerospike Graph and Aerospike Vector Search are distinct product surfaces in current public documentation.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Aerospike-centric operational graphs with separate vector services if needed",
      "profile": "databases/aerospike/README.md",
      "source_urls": [
        "https://aerospike.com/docs/graph/quick-start",
        "https://aerospike.com/docs/graph/manage/summary/",
        "https://aerospike.com/docs/vector/quick-start"
      ],
      "notes": "The repo previously overstated native graph-plus-vector coupling; the public docs show separate graph and vector product lines."
    },
    {
      "id": "tarantool-graph",
      "name": "Tarantool Graph DB",
      "model": "property",
      "status": "emerging",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.tarantool.io/en/solutions/vector-database/",
      "query_interfaces": ["openCypher", "APIs"],
      "vector_strategy": "native",
      "vector_summary": "Tarantool markets a graph-vector database product, but the public material still reads as early-stage product collateral rather than mature documentation.",
      "ha": "Claimed",
      "deployment": "Self-hosted",
      "best_fit": "Teams specifically evaluating Tarantool's graph-vector direction",
      "profile": "databases/tarantool-graph/README.md",
      "source_urls": [
        "https://www.tarantool.io/en/solutions/vector-database/"
      ],
      "notes": "Keep in an emerging bucket until public docs and production references become deeper."
    },
    {
      "id": "sparksee",
      "name": "Sparksee",
      "model": "specialized",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.sparsity-technologies.com/sparksee-graph-database-management-system/",
      "query_interfaces": ["Embedded APIs"],
      "vector_strategy": "none",
      "vector_summary": "Sparksee public docs describe an embedded graph database API, not native vector search.",
      "ha": "No",
      "deployment": "Embedded",
      "best_fit": "Embedded and mobile graph applications",
      "profile": "databases/sparksee/README.md",
      "source_urls": [
        "https://www.sparsity-technologies.com/sparksee-graph-database-management-system/",
        "https://www.sparsity-technologies.com/doc-sparksee/UserManual/html/GraphDatabase.html"
      ],
      "notes": "Specialized embedded graph database, not a modern GraphRAG platform."
    },
    {
      "id": "tugraph",
      "name": "TuGraph",
      "model": "property",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://tugraph-db.readthedocs.io",
      "query_interfaces": ["openCypher", "GQL", "REST", "Procedures"],
      "vector_strategy": "none",
      "vector_summary": "Current public docs describe property-graph storage and GQL/openCypher, but not native vector search.",
      "ha": "Yes",
      "deployment": "Self-hosted",
      "best_fit": "Property graph plus graph-computing workloads without native vector requirements",
      "profile": "databases/tugraph/README.md",
      "source_urls": [
        "https://tugraph-db.readthedocs.io/en/v4.3.0/3.quick-start/1.preparation.html",
        "https://tugraph-db.readthedocs.io/en/v4.0.1/5.developer-manual/6.interface/1.query/2.gql/1.intro.html"
      ],
      "notes": "Strong missing LPG entry from the previous repo version; now added to active scope."
    },
    {
      "id": "agensgraph",
      "name": "AgensGraph",
      "model": "multi",
      "status": "production",
      "license": "Commercial community download / enterprise",
      "license_type": "source-available",
      "website": "https://www.bitnineglobal.com",
      "query_interfaces": ["openCypher", "SQL"],
      "vector_strategy": "none",
      "vector_summary": "Current public AgensGraph materials focus on PostgreSQL plus graph, not vector search.",
      "ha": "PostgreSQL dependent",
      "deployment": "Self-hosted",
      "best_fit": "PostgreSQL-compatible graph workloads using a Bitnine stack",
      "profile": "databases/agensgraph/README.md",
      "source_urls": [
        "https://www.bitnineglobal.com/",
        "https://www.bitnineglobal.com/download"
      ],
      "notes": "Added because it is still a relevant PostgreSQL-plus-graph product in the LPG landscape."
    },
    {
      "id": "fluree",
      "name": "Fluree",
      "model": "multi",
      "status": "production",
      "license": "Commercial plus open components",
      "license_type": "source-available",
      "website": "https://flur.ee",
      "query_interfaces": ["JSON-LD query", "SPARQL", "GraphQL"],
      "vector_strategy": "unclear",
      "vector_summary": "Fluree clearly positions itself for GraphRAG and semantic AI, but current public technical docs do not expose a clearly documented general-purpose ANN vector subsystem.",
      "ha": "Yes",
      "deployment": "Self-hosted, managed cloud",
      "best_fit": "Semantic layer and JSON-LD knowledge-graph workflows",
      "profile": "databases/fluree/README.md",
      "source_urls": [
        "https://flur.ee/",
        "https://developers.flur.ee/docs/reference/full-text-search/",
        "https://developers.flur.ee/docs/reference/querying/advanced-features/"
      ],
      "notes": "The previous repo version overstated vector specifics beyond what current public docs make explicit."
    },
    {
      "id": "redisgraph",
      "name": "RedisGraph",
      "model": "property",
      "status": "legacy",
      "license": "Source available",
      "license_type": "source-available",
      "website": "https://github.com/RedisGraph/RedisGraph",
      "query_interfaces": ["Cypher"],
      "vector_strategy": "none",
      "vector_summary": "No native vector graph search; the project is archived and replaced in practice by FalkorDB.",
      "ha": "Historical",
      "deployment": "Self-hosted",
      "best_fit": "Legacy migration reference",
      "profile": "databases/redisgraph/README.md",
      "source_urls": [
        "https://github.com/RedisGraph/RedisGraph"
      ],
      "notes": "Kept only to document the historical path toward FalkorDB."
    },
    {
      "id": "cozo",
      "name": "CozoDB",
      "model": "multi",
      "status": "production",
      "license": "MPL-2.0",
      "license_type": "permissive",
      "website": "https://www.cozodb.org",
      "query_interfaces": ["CozoScript (Datalog)"],
      "vector_strategy": "native",
      "vector_summary": "HNSW vector indexes are a first-class index type in CozoScript and can be combined with Datalog graph traversals in a single query.",
      "ha": "Embedded or single-node server",
      "deployment": "Embedded library, single-node server, or browser via WASM",
      "best_fit": "Embedded knowledge-graph and graph-plus-vector workloads with a Datalog query surface",
      "profile": "databases/cozo/README.md",
      "source_urls": [
        "https://github.com/cozodb/cozo",
        "https://docs.cozodb.org/en/latest/vector.html"
      ],
      "notes": "One of the few open-source graph stores with a genuinely native HNSW index baked into the query engine."
    },
    {
      "id": "oracle-graph",
      "name": "Oracle Graph",
      "model": "multi",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.oracle.com/database/graph/",
      "query_interfaces": ["PGQL", "SPARQL 1.1", "SQL/PGQ", "Gremlin"],
      "vector_strategy": "integrated",
      "vector_summary": "Oracle Database 23ai bundles AI Vector Search alongside PGQL and RDF, so graph traversals and vector similarity can be composed in a single query.",
      "ha": "Enterprise RAC and Data Guard",
      "deployment": "Oracle Database 23ai (self-hosted, cloud, or 23ai Free)",
      "best_fit": "Enterprise workloads that need dual RDF and property-graph with native vector search in the same engine",
      "profile": "databases/oracle-graph/README.md",
      "source_urls": [
        "https://docs.oracle.com/en/database/oracle/property-graph/",
        "https://www.oracle.com/database/graph/"
      ],
      "notes": "Historically two separate layers (Oracle Spatial and Graph, Oracle RDF Graph); now unified under the Oracle Graph brand inside Oracle Database 23ai."
    },
    {
      "id": "spanner-graph",
      "name": "Google Spanner Graph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://cloud.google.com/spanner/docs/graph/overview",
      "query_interfaces": ["GQL (ISO/IEC 39075)", "SQL"],
      "vector_strategy": "integrated",
      "vector_summary": "Spanner exposes graph queries in ISO GQL alongside vector similarity functions and ScaNN-backed indexes in the same distributed SQL engine.",
      "ha": "Global, strongly consistent, multi-region",
      "deployment": "Google Cloud managed service",
      "best_fit": "Globally distributed property-graph workloads that need transactional consistency plus vector search",
      "profile": "databases/spanner-graph/README.md",
      "source_urls": [
        "https://cloud.google.com/spanner/docs/graph/overview",
        "https://cloud.google.com/spanner/docs/find-approximate-nearest-neighbors"
      ],
      "notes": "First hyperscaler graph offering built on the ISO GQL standard; vector search and graph share one storage layer."
    },
    {
      "id": "gel",
      "name": "Gel (formerly EdgeDB)",
      "model": "specialized",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://www.geldata.com",
      "query_interfaces": ["EdgeQL", "GraphQL", "SQL (read-only)"],
      "vector_strategy": "integrated",
      "vector_summary": "Gel ships an official vector extension (ext::pgvector) that exposes ANN search from EdgeQL over its graph-relational object model.",
      "ha": "Replication via Postgres backend",
      "deployment": "Self-hosted, Docker, or Gel Cloud",
      "best_fit": "Applications that want a graph-relational object model with first-class vector search",
      "profile": "databases/gel/README.md",
      "source_urls": [
        "https://docs.geldata.com/",
        "https://www.geldata.com/blog/edgedb-is-now-gel-and-postgres-is-the-future"
      ],
      "notes": "EdgeDB rebranded to Gel in 2025. The core model is graph-relational, not classical RDF or LPG, so it is catalogued as specialized."
    },
    {
      "id": "datomic",
      "name": "Datomic",
      "model": "specialized",
      "status": "production",
      "license": "Apache 2.0 (binaries)",
      "license_type": "source-available",
      "website": "https://www.datomic.com",
      "query_interfaces": ["Datalog", "Pull API", "Entity API"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; Datomic is an immutable entity-attribute-value store queried with Datalog.",
      "ha": "Datomic Cloud on AWS or Datomic Pro with external storage",
      "deployment": "Self-hosted (Datomic Pro) or managed on AWS (Datomic Cloud)",
      "best_fit": "Immutable entity-attribute-value knowledge bases with time-travel and Datalog",
      "profile": "databases/datomic/README.md",
      "source_urls": [
        "https://docs.datomic.com/",
        "https://blog.datomic.com/2023/04/datomic-is-free.html"
      ],
      "notes": "Datomic Pro binaries were released under Apache 2.0 and made free of licensing fees in April 2023 by Nubank. Source code is not public, so the license family is marked as source-available."
    },
    {
      "id": "hypergraphdb",
      "name": "HyperGraphDB",
      "model": "specialized",
      "status": "legacy",
      "license": "LGPL-3.0",
      "license_type": "copyleft",
      "website": "https://www.hypergraphdb.org",
      "query_interfaces": ["Java API"],
      "vector_strategy": "none",
      "vector_summary": "No vector search capability; HyperGraphDB is a general-purpose hypergraph store where edges can connect arbitrary numbers of nodes.",
      "ha": "None",
      "deployment": "Embedded Java library",
      "best_fit": "Research or legacy systems that specifically need a true hypergraph model",
      "profile": "databases/hypergraphdb/README.md",
      "source_urls": [
        "https://www.hypergraphdb.org",
        "https://github.com/hypergraphdb/hypergraphdb"
      ],
      "notes": "Included as the reference hypergraph anchor. Activity has slowed significantly since 2017."
    },
    {
      "id": "anzograph",
      "name": "AnzoGraph DB",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.cambridgesemantics.com/anzograph/",
      "query_interfaces": ["SPARQL 1.1", "openCypher"],
      "vector_strategy": "unclear",
      "vector_summary": "Public documentation covers large-scale SPARQL, reasoning, and analytics; no general-purpose vector index is currently documented.",
      "ha": "Massively parallel in-memory cluster",
      "deployment": "Self-hosted, Kubernetes, or Cambridge Semantics Anzo platform",
      "best_fit": "Enterprise MPP SPARQL analytics and knowledge-graph workloads",
      "profile": "databases/anzograph/README.md",
      "source_urls": [
        "https://docs.cambridgesemantics.com/anzograph/",
        "https://www.cambridgesemantics.com/anzograph/"
      ],
      "notes": "AnzoGraph DB is the MPP SPARQL engine embedded inside the broader Anzo platform from Cambridge Semantics; also marketed as Altair Graph Studio after the Altair acquisition."
    },
    {
      "id": "weaviate",
      "name": "Weaviate",
      "model": "specialized",
      "status": "production",
      "license": "BSD-3-Clause",
      "license_type": "permissive",
      "website": "https://weaviate.io",
      "query_interfaces": ["GraphQL", "REST", "gRPC"],
      "vector_strategy": "native",
      "vector_summary": "Weaviate is a vector-first database with HNSW indexes and typed cross-references between objects that form a navigable knowledge graph.",
      "ha": "Sharding and replication across cluster nodes",
      "deployment": "Self-hosted, Docker, Kubernetes, or Weaviate Cloud",
      "best_fit": "Vector-first knowledge graphs with GraphQL access and schema-defined cross-references",
      "profile": "databases/weaviate/README.md",
      "source_urls": [
        "https://github.com/weaviate/weaviate",
        "https://docs.weaviate.io/weaviate"
      ],
      "notes": "Primarily a vector database; the graph dimension comes from typed cross-references between objects rather than arbitrary edges. Catalogued as specialized because it is not a classical LPG or RDF engine."
    },
    {
      "id": "cayley",
      "name": "Cayley",
      "model": "multi",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://cayley.io",
      "query_interfaces": ["Gizmo (JavaScript DSL)", "GraphQL", "MQL"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; Cayley is a linked-data graph database inspired by the Google Knowledge Graph and Freebase with pluggable backend stores.",
      "ha": "Depends on the chosen backend store",
      "deployment": "Self-hosted with LevelDB, BoltDB, SQL, PostgreSQL, or MongoDB backends",
      "best_fit": "Open-source linked-data projects that want a Go-native engine with multiple query languages and pluggable storage",
      "profile": "databases/cayley/README.md",
      "source_urls": [
        "https://github.com/cayleygraph/cayley",
        "https://cayley.io"
      ],
      "notes": "Google-origin project inspired by the Freebase / Knowledge Graph architecture. Maintained by the community under cayleygraph/cayley on GitHub."
    },
    {
      "id": "dse-graph",
      "name": "DataStax Enterprise Graph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.datastax.com/products/datastax-graph",
      "query_interfaces": ["Gremlin", "CQL"],
      "vector_strategy": "none",
      "vector_summary": "No vector capability is documented in the DSE Graph engine; vector features live in DataStax's separate Astra DB product line.",
      "ha": "Cassandra-native multi-datacenter replication",
      "deployment": "Self-hosted DataStax Enterprise cluster",
      "best_fit": "Enterprise graph workloads backed by Apache Cassandra's distributed storage and linear scale-out",
      "profile": "databases/dse-graph/README.md",
      "source_urls": [
        "https://docs.datastax.com/en/dse/6.9/graph/graph-contents.html",
        "https://www.datastax.com/products/datastax-graph"
      ],
      "notes": "Built on top of Apache Cassandra inside DataStax Enterprise. Classic DSE Graph and the newer DataStax Graph engine share the same product surface from DSE 6.8 onward."
    },
    {
      "id": "rdf4j",
      "name": "Eclipse RDF4J",
      "model": "rdf",
      "status": "production",
      "license": "EDL 1.0 / EPL 2.0",
      "license_type": "permissive",
      "website": "https://rdf4j.org",
      "query_interfaces": ["SPARQL 1.1", "SeRQL"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; RDF4J is a Java framework whose bundled Native Store and Memory Store provide embeddable SPARQL 1.1 endpoints.",
      "ha": "Single-node stores; clustering depends on the SAIL backend",
      "deployment": "Embedded Java library, standalone RDF4J Server, or Workbench",
      "best_fit": "Java applications that need an embeddable SPARQL 1.1 store or a framework for building one",
      "profile": "databases/rdf4j/README.md",
      "source_urls": [
        "https://rdf4j.org/about/",
        "https://projects.eclipse.org/projects/technology.rdf4j"
      ],
      "notes": "Formerly known as OpenRDF Sesame. Ships both an on-disk (Native Store) and in-memory (Memory Store) SAIL implementation, plus a pluggable SAIL architecture for third-party backends."
    },
    {
      "id": "atomic-server",
      "name": "AtomicServer",
      "model": "specialized",
      "status": "emerging",
      "license": "MIT",
      "license_type": "permissive",
      "website": "https://atomicserver.eu",
      "query_interfaces": ["REST", "Atomic queries"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; AtomicServer is the reference implementation of the Atomic Data specification, a type-safe, RDF-inspired, JSON-compatible graph data model.",
      "ha": "Single-node (sled-backed storage)",
      "deployment": "Self-hosted, single binary",
      "best_fit": "Headless CMS and real-time collaborative data apps that want a type-safe, RDF-inspired graph model",
      "profile": "databases/atomic-server/README.md",
      "source_urls": [
        "https://docs.atomicdata.dev/",
        "https://github.com/atomicdata-dev/atomic-server"
      ],
      "notes": "Reference implementation of the Atomic Data specification. The underlying data model combines RDF-style links with JSON ergonomics and runtime type safety."
    },
    {
      "id": "helixdb",
      "name": "HelixDB",
      "model": "multi",
      "status": "emerging",
      "license": "AGPL-3.0",
      "license_type": "copyleft",
      "website": "https://www.helix-db.com",
      "query_interfaces": ["HelixQL"],
      "vector_strategy": "native",
      "vector_summary": "HelixDB is a Rust-native graph-vector database that ships both graph traversal and vector search as first-class primitives inside HelixQL, its strongly-typed compiled query language.",
      "ha": "Single-node (LMDB-backed storage)",
      "deployment": "Self-hosted; managed service available for selected users",
      "best_fit": "RAG and AI applications that need graph traversal and vector search in a single typed query surface",
      "profile": "databases/helixdb/README.md",
      "source_urls": [
        "https://github.com/HelixDB/helix-db",
        "https://www.helix-db.com"
      ],
      "notes": "Rust implementation with LMDB storage. AGPL-3.0 has implications for SaaS distribution; review the license before embedding in commercial services."
    },
    {
      "id": "dydra",
      "name": "Dydra",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://dydra.com",
      "query_interfaces": ["SPARQL 1.1", "GraphQL", "Linked Data Platform"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; Dydra is a versioned cloud RDF store with first-class temporal snapshots and SPARQL 1.1 support.",
      "ha": "Managed cloud (recent work integrates the RonDB distributed backend)",
      "deployment": "Managed cloud service operated by Datagraph",
      "best_fit": "Cloud-native RDF applications that need versioned, time-travelable SPARQL stores",
      "profile": "databases/dydra/README.md",
      "source_urls": [
        "https://docs.dydra.com/dydra",
        "https://docs.dydra.com/sparql"
      ],
      "notes": "A 'Git for graphs' approach — every store state is fully addressable via a REVISION argument in SPARQL. Recent work with Hopsworks/RonDB targets trillion-triple scale."
    },
    {
      "id": "alibaba-gdb",
      "name": "Alibaba Cloud Graph Database",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.alibabacloud.com/product/graph-database",
      "query_interfaces": ["Gremlin", "openCypher"],
      "vector_strategy": "none",
      "vector_summary": "No native vector capability is documented in the GDB engine; vector workloads live in Alibaba Cloud's separate Lindorm and AnalyticDB product lines.",
      "ha": "Managed multi-AZ deployment",
      "deployment": "Alibaba Cloud managed service",
      "best_fit": "Alibaba Cloud workloads that need a real-time property graph with TinkerPop-compatible tooling",
      "profile": "databases/alibaba-gdb/README.md",
      "source_urls": [
        "https://www.alibabacloud.com/help/product/102714.htm",
        "https://www.alibabacloud.com/product/graph-database"
      ],
      "notes": "Alibaba Cloud's managed property graph. Supports the Apache TinkerPop Gremlin query language as the primary surface; listed as an official TinkerPop provider."
    },
    {
      "id": "ibm-db2-graph",
      "name": "IBM Db2 Graph",
      "model": "multi",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.ibm.com/docs/en/db2/11.5?topic=da-db2-graph",
      "query_interfaces": ["Gremlin"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index in the graph layer; Db2 Graph is a virtual graph overlay on Db2 relational tables with no documented vector integration.",
      "ha": "Db2-native HADR and pureScale",
      "deployment": "Included with IBM Db2 (self-hosted or Cloud Pak for Data)",
      "best_fit": "Organizations already on IBM Db2 that want a virtual property graph over existing relational data without a separate engine",
      "profile": "databases/ibm-db2-graph/README.md",
      "source_urls": [
        "https://www.ibm.com/docs/en/db2/11.5?topic=da-db2-graph",
        "https://www.ibm.com/docs/en/cloud-paks/cp-data/4.0?topic=applications-db2-graph"
      ],
      "notes": "Db2 Graph transforms Gremlin queries into SQL against a Db2 database via a Graph overlay file that maps rows to vertices and edges. Listed as an official Apache TinkerPop provider."
    },
    {
      "id": "huawei-ges",
      "name": "Huawei Graph Engine Service",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.huaweicloud.com/intl/en-us/product/ges.html",
      "query_interfaces": ["Gremlin", "Cypher"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented in GES; Huawei Cloud exposes vector search through separate services rather than inside the graph engine.",
      "ha": "Managed distributed cluster",
      "deployment": "Huawei Cloud managed service",
      "best_fit": "Huawei Cloud workloads that need a fully managed distributed property graph at tens of billions of vertices",
      "profile": "databases/huawei-ges/README.md",
      "source_urls": [
        "https://support.huaweicloud.com/intl/en-us/productdesc-ges/ges_04_0001.html",
        "https://www.huaweicloud.com/intl/en-us/product/ges.html"
      ],
      "notes": "Backed by Huawei's in-house EYWA graph kernel. Supports Apache TinkerPop Gremlin 3.4 and a Cypher surface. Listed as an official Apache TinkerPop provider."
    },
    {
      "id": "ongdb",
      "name": "ONgDB",
      "model": "property",
      "status": "production",
      "license": "GPL-3.0 / AGPL-3.0",
      "license_type": "copyleft",
      "website": "https://graphfoundation.org/ongdb/",
      "query_interfaces": ["Cypher"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; ONgDB is a community fork of Neo4j Enterprise 3.4 that predates the HNSW vector work in upstream Neo4j.",
      "ha": "Causal clustering (inherited from Neo4j Enterprise 3.4)",
      "deployment": "Self-hosted",
      "best_fit": "Teams that want a fully open-source continuation of Neo4j Enterprise without the open-core split",
      "profile": "databases/ongdb/README.md",
      "source_urls": [
        "https://github.com/graphfoundation/ongdb",
        "https://graphfoundation.org/ongdb/"
      ],
      "notes": "ONgDB (Open Native Graph Database) is an independent fork of Neo4j Enterprise Edition 3.4 maintained by the Graph Foundation. Community distributions are GPL-3.0 and Enterprise distributions are AGPL-3.0. Not affiliated with Neo4j, Inc."
    },
    {
      "id": "ontop",
      "name": "Ontop",
      "model": "rdf",
      "status": "production",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://ontop-vkg.org",
      "query_interfaces": ["SPARQL 1.1"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; Ontop is a virtual knowledge graph system that translates SPARQL into SQL on top of a relational database.",
      "ha": "Inherits availability from the backing relational database",
      "deployment": "Self-hosted SPARQL endpoint over an existing relational database",
      "best_fit": "Organizations that want to expose an existing relational database as a virtual RDF knowledge graph without materializing triples",
      "profile": "databases/ontop/README.md",
      "source_urls": [
        "https://github.com/ontop/ontop",
        "https://ontop-vkg.org/guide/"
      ],
      "notes": "Implements the Ontology-Based Data Access (OBDA) pattern. Uses R2RML or Ontop's native OBDA mappings to translate SPARQL queries into SQL at runtime, so data stays in the source relational store."
    },
    {
      "id": "marmotta",
      "name": "Apache Marmotta",
      "model": "rdf",
      "status": "legacy",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://marmotta.apache.org",
      "query_interfaces": ["SPARQL 1.1", "Linked Data Platform"],
      "vector_strategy": "none",
      "vector_summary": "No vector capability; Apache Marmotta was retired to the Apache Attic in November 2020 and is no longer actively maintained.",
      "ha": "None",
      "deployment": "Self-hosted Java web application",
      "best_fit": "Historical reference for W3C Linked Data Platform deployments",
      "profile": "databases/marmotta/README.md",
      "source_urls": [
        "https://marmotta.apache.org",
        "https://en.wikipedia.org/wiki/Apache_Marmotta"
      ],
      "notes": "Retired to the Apache Attic in November 2020. Included as a historical reference for the W3C Linked Data Platform recommendation it helped implement."
    },
    {
      "id": "relationalai",
      "name": "RelationalAI",
      "model": "specialized",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://relational.ai",
      "query_interfaces": ["Rel (Datalog-based)"],
      "vector_strategy": "unclear",
      "vector_summary": "Rel is a declarative relational-knowledge-graph language; current public documentation does not expose a clear general-purpose vector index surface.",
      "ha": "Snowflake-native managed service",
      "deployment": "Managed co-processor on Snowflake (Snowflake Marketplace)",
      "best_fit": "Enterprises on Snowflake that want a declarative relational knowledge graph layer with Datalog-style recursion",
      "profile": "databases/relationalai/README.md",
      "source_urls": [
        "https://www.relational.ai/",
        "https://docs.relational.ai/rel/concepts/relational-knowledge-graphs"
      ],
      "notes": "Ships a relational knowledge graph management system built around Rel, a Datalog-inspired query language. Available as a co-processor to Snowflake through the Snowflake Marketplace."
    },
    {
      "id": "raphtory",
      "name": "Raphtory",
      "model": "specialized",
      "status": "emerging",
      "license": "GPL-3.0",
      "license_type": "copyleft",
      "website": "https://www.pometry.com",
      "query_interfaces": ["Python API", "GraphQL"],
      "vector_strategy": "none",
      "vector_summary": "No general-purpose vector index; Raphtory focuses on temporal analytics with time-travel, multilayer modelling, and motif detection rather than similarity search.",
      "ha": "Single-node embedded or server (out-of-memory scaling via subscription)",
      "deployment": "Embedded Rust/Python library or standalone GraphQL server",
      "best_fit": "Temporal graph analytics workloads that need time-travel, multilayer modelling, and dynamic scoring",
      "profile": "databases/raphtory/README.md",
      "source_urls": [
        "https://github.com/Pometry/Raphtory",
        "https://docs.pometry.com/"
      ],
      "notes": "Developed by Pometry. In-memory vectorized temporal graph engine written in Rust with Python bindings. A managed on-disk scaling path is available under subscription."
    },
    {
      "id": "katana-graph",
      "name": "Katana Graph",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://katanagraph.ai",
      "query_interfaces": ["Cypher", "Gremlin", "SPARQL", "Python API"],
      "vector_strategy": "unclear",
      "vector_summary": "Public documentation emphasizes graph analytics and ML pipelines; a general-purpose vector index is not clearly exposed in public materials.",
      "ha": "Distributed cluster across x86, Optane, GPU nodes",
      "deployment": "Self-hosted on clusters, Intel Optane, or multi-GPU systems",
      "best_fit": "High-performance distributed graph analytics on very large graphs with heterogeneous hardware",
      "profile": "databases/katana-graph/README.md",
      "source_urls": [
        "https://katanagraph.ai/",
        "https://katanagraph.ai/why-katana"
      ],
      "notes": "Founded in 2020 by researchers from the University of Texas at Austin. Targets large-scale graph intelligence workloads in security, finance, and healthcare; backed by Intel Capital."
    },
    {
      "id": "trovares-xgt",
      "name": "Trovares xGT (Rocketgraph)",
      "model": "property",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://rocketgraph.com",
      "query_interfaces": ["Python API", "openCypher"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; xGT is a high-performance in-memory property graph engine focused on analytic query throughput.",
      "ha": "Single-user in-memory engine",
      "deployment": "Self-hosted on large-memory servers",
      "best_fit": "Single-user analytic graph workloads on very large in-memory graphs",
      "profile": "databases/trovares-xgt/README.md",
      "source_urls": [
        "https://docs.trovares.com/",
        "https://rocketgraph.com/"
      ],
      "notes": "Developed by Trovares (now branded Rocketgraph). Founded in 2014 to commercialize parallel graph work from the US Department of Defense and Pacific Northwest National Laboratory."
    },
    {
      "id": "s2graph",
      "name": "Apache S2Graph",
      "model": "property",
      "status": "legacy",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://s2graph.apache.org",
      "query_interfaces": ["Gremlin", "REST"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; S2Graph is a TinkerPop-compatible OLTP graph layer on top of Apache HBase.",
      "ha": "Inherits HBase distribution and replication",
      "deployment": "Self-hosted on an Apache HBase cluster",
      "best_fit": "Organizations already running HBase that need a TinkerPop-compatible graph layer",
      "profile": "databases/s2graph/README.md",
      "source_urls": [
        "https://s2graph.apache.org",
        "https://github.com/apache/incubator-s2graph"
      ],
      "notes": "Still in Apache Incubation; community activity has been light. Listed as an official Apache TinkerPop provider."
    },
    {
      "id": "poolparty",
      "name": "PoolParty Semantic Suite",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.poolparty.biz",
      "query_interfaces": ["SPARQL 1.1", "SKOS", "REST"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is exposed at the store level; PoolParty focuses on SKOS thesaurus management and semantic search over an RDF graph store.",
      "ha": "Deployment-dependent",
      "deployment": "Self-hosted or private cloud suite",
      "best_fit": "Enterprise taxonomy, controlled vocabulary, and semantic search over W3C standards",
      "profile": "databases/poolparty/README.md",
      "source_urls": [
        "https://www.poolparty.biz/semantic-integrator",
        "https://en.wikipedia.org/wiki/PoolParty_Semantic_Suite"
      ],
      "notes": "Developed by the Semantic Web Company in Vienna. Built around SKOS and SKOS-XL for enterprise taxonomies, ontologies, and knowledge graphs."
    },
    {
      "id": "topbraid-edg",
      "name": "TopBraid EDG",
      "model": "rdf",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://www.topquadrant.com/topbraid-edg/",
      "query_interfaces": ["SPARQL", "SHACL", "GraphQL"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is exposed; TopBraid EDG focuses on RDF knowledge graphs, SHACL validation, and SPARQL or GraphQL access to governed data.",
      "ha": "Deployment-dependent",
      "deployment": "Self-hosted or private cloud",
      "best_fit": "Enterprise data governance, business glossary, and reference data management backed by an RDF knowledge graph",
      "profile": "databases/topbraid-edg/README.md",
      "source_urls": [
        "https://www.topquadrant.com/topbraid-edg/",
        "https://www.topquadrant.com/doc/latest/introduction/index.html"
      ],
      "notes": "TopBraid Enterprise Data Governance (EDG) is TopQuadrant's flagship platform. Uses SHACL alongside SPARQL and GraphQL for governed knowledge graph applications."
    },
    {
      "id": "indradb",
      "name": "IndraDB",
      "model": "property",
      "status": "production",
      "license": "MPL-2.0",
      "license_type": "permissive",
      "website": "https://indradb.github.io",
      "query_interfaces": ["gRPC", "Rust library", "CLI"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index is documented; IndraDB is a directed typed graph store inspired by Facebook's TAO datastore.",
      "ha": "Pluggable datastores (in-memory, sled, RocksDB, PostgreSQL)",
      "deployment": "Embedded Rust library, standalone server, or CLI",
      "best_fit": "Embedded or cross-language property-graph workloads that need a minimal Rust-native engine",
      "profile": "databases/indradb/README.md",
      "source_urls": [
        "https://github.com/indradb/indradb",
        "https://indradb.github.io"
      ],
      "notes": "Design is inspired by Facebook's TAO graph datastore. Ships a server (gRPC) and an embeddable library, with pluggable underlying storage."
    },
    {
      "id": "hgraphdb",
      "name": "HGraphDB",
      "model": "property",
      "status": "legacy",
      "license": "Apache 2.0",
      "license_type": "permissive",
      "website": "https://github.com/rayokota/hgraphdb",
      "query_interfaces": ["Gremlin"],
      "vector_strategy": "none",
      "vector_summary": "No native vector index; HGraphDB is a TinkerPop 3 client layer that turns Apache HBase into a graph database.",
      "ha": "Inherits HBase distribution and replication",
      "deployment": "Client library on top of an Apache HBase cluster",
      "best_fit": "Organizations already running HBase that need a TinkerPop 3 graph layer with Giraph or Spark GraphFrames integration",
      "profile": "databases/hgraphdb/README.md",
      "source_urls": [
        "https://github.com/rayokota/hgraphdb",
        "https://blogs.apache.org/hbase/entry/hgraphdb_hbase_as_a_tinkerpop"
      ],
      "notes": "Created by Robert Yokota. Provides Gremlin over HBase via the TinkerPop 3 API. Listed as an official Apache TinkerPop provider."
    },
    {
      "id": "sap-hana-graph",
      "name": "SAP HANA Graph",
      "model": "multi",
      "status": "production",
      "license": "Commercial",
      "license_type": "commercial",
      "website": "https://help.sap.com/docs/hana-cloud-database/sap-hana-cloud-sap-hana-database-property-graph-engine-reference/",
      "query_interfaces": ["openCypher", "GraphScript", "SQL"],
      "vector_strategy": "integrated",
      "vector_summary": "SAP HANA Cloud exposes a dedicated vector engine alongside the property graph engine in the same database, so graph pattern matching and vector similarity can be combined in a single SQL session.",
      "ha": "SAP HANA system replication and scale-out",
      "deployment": "SAP HANA Cloud (managed) or self-hosted SAP HANA",
      "best_fit": "Organizations already standardized on SAP HANA that want property-graph analytics and vector search without a second system",
      "profile": "databases/sap-hana-graph/README.md",
      "source_urls": [
        "https://help.sap.com/docs/hana-cloud-database/sap-hana-cloud-sap-hana-database-property-graph-engine-reference/appendix-d-opencypher-cheat-sheet",
        "https://developers.sap.com/group.hana-aa-graph-overview.html"
      ],
      "notes": "SAP HANA's native graph engine ships as part of SAP HANA core functionality. Supports openCypher for pattern matching and GraphScript for implementing graph algorithms inside the database."
    }
  ]
}