Tentris

Tentris is an RDF database for teams that want to keep graph data in their own infrastructure and query it with SPARQL 1.1. Its strongest fit is non-commercial development and research; production users should weigh the paid, per-core licensing against their storage and support needs.

Overview

Tentris stores RDF knowledge graphs in a single compressed Hypertrie index designed to reduce memory use. That design is relevant when graph workloads put pressure on memory, though the Community plan’s in-memory-only storage limits its usefulness for deployments that need persistence.

Query execution combines adaptive planning with worst-case optimal joins, streams results, and avoids materializing intermediate results. Those choices are aimed at keeping query processing efficient as graph workloads grow. Tentris supports RDF 1.1 and SPARQL 1.1, so an existing RDF data model need not change when switching. Federated queries are supported, but reasoning is not.

Tentris is also positioned as a knowledge layer for GraphRAG, AI agents, and semantic applications, where deterministic retrieval over connected knowledge is useful. The company says its benchmark figures are tied to reproducible datasets, queries, hardware, and methods, giving teams a basis for judging published results against their own workloads. The database grew from more than a decade of research at Paderborn University, and Tentris says it builds the product in Germany.

Key features

RDF storage and query execution

The compressed index, streaming results, and avoidance of intermediate-result materialization make Tentris worth considering for memory-conscious graph query workloads. These are design goals, not a substitute for sizing a deployment against its own data and query patterns.

Formats and deployment

Tentris accepts N-Triples, Turtle, N-Quads, TriG, and RDF/XML. The software is self-hosted, with Linux, macOS, Docker, and Python installation options and x86-64 and ARM AARCH64 support. That flexibility helps teams choose an environment, but puts deployment and operations in their hands.

Data control and AI use

The EULA says the downloadable software runs within the licensee’s infrastructure and Tentris does not host, receive, or otherwise access its content. That is a useful fit for organizations that need to keep data under their own control. GraphRAG and agent use cases are part of Tentris’s positioning, but teams requiring built-in reasoning should look elsewhere.

Pricing

Tentris uses a freemium model. The free Community plan costs nothing and includes a single-core license, in-memory storage, community support, SPARQL 1.1 and RDF 1.1 support, and transactions. It suits developers, students, and researchers evaluating or building non-commercial projects. The EULA limits non-commercial use to private, academic, educational, scientific, evaluation, or research purposes and excludes production and revenue-generating use. Although community support is included, Tentris provides no support or maintenance for this software under the EULA.

The Commercial plan has custom pricing, billed Annual, and is licensed per CPU core starting at 12 cores. It adds persistent storage, ACID transactions, an optional commercial SLA, and a 60-day trial license. This is the relevant path for production workloads or organizations that need commercial terms, but the minimum core count and per-core licensing make it important to match the deployment size to budget before committing.

The Cloud plan has custom pricing, billed Pay per Use, and is marked coming soon. Its proposed offer includes Azure, AWS, and Google Cloud, daily backups, configurable retention, on-demand scaling, and an SLA agreement. It may suit teams that want a managed deployment, but it is not yet an option to plan around.

Platforms

Tentris runs on Linux and macOS and supports self-hosted deployment. Docker and Python installation options and x86-64 and ARM AARCH64 support broaden its fit across infrastructure choices. The planned cloud service names Azure, AWS, and Google Cloud, but is coming soon.

Who it's for

Community is best for non-commercial experimentation, academic work, and research where single-core, in-memory operation is sufficient. Commercial is aimed at production and organizations with data-sovereignty or strict-compliance needs, especially where persistent storage and an optional SLA matter. Teams that need reasoning, or that cannot accept self-hosting and per-core commercial licensing, should compare other RDF databases.

Pros and cons

Pros: RDF 1.1 and SPARQL 1.1 compliance avoids a required data-model change; federated queries and five RDF formats support varied graph workflows.

Pros: A compressed index and streaming query design target lower memory use, while reproducible benchmark materials give teams a way to assess published performance claims.

Pros: Self-hosting keeps content within the licensee’s infrastructure, an advantage for data control.

Cons: Community is restricted to non-commercial use, single-core licensing, and in-memory storage, ruling it out for production and persistent workloads.

Cons: Commercial starts at 12 CPU cores and has custom pricing, which may be a poor fit for smaller deployments.

Cons: Reasoning is not supported, and the cloud offering is not yet available.

Alternatives

For a wider comparison, browse RDF Databases. Choose RDFox if its 30-day non-commercial evaluation license better matches a short assessment period; commercial pricing is available by quote. Fluree DB is an alternative for teams seeking an open-source database with native JSON-LD and SPARQL that can run locally or self-host.

GraphDB may suit users who want its free edition’s Lucene connector, five-repository allowance, and one parallel query limit. Choose QLever if an Apache 2.0 open-source RDF database is the priority. Stardog offers a renewable one-year free license capped at 1 million stored edges and three databases, making it a different option for bounded free use.

Apache Jena Fuseki is a free, self-hosted server under Apache License 2.0. Eclipse RDF4J is a free open-source Java framework, with its latest release requiring Java 25 or newer. Virtuoso Universal Server is another option to compare for RDF database needs.

Verdict

Tentris is a strong candidate for developers and researchers who want self-hosted SPARQL 1.1 graph storage, federated queries, and a non-commercial route to evaluate the system. Its compact-index and streaming-execution design is compelling, but Community’s license and in-memory limits sharply constrain use beyond evaluation and research. Production teams should choose it only if persistent storage, the 12-core commercial minimum, and custom per-core terms fit; otherwise, compare alternatives with a better match for their deployment and licensing needs.