DEV.co
Vector Databases · endee-io

endee

Endee is an open-source vector database written in C++ designed to store and search up to 1 billion vectors on a single node, supporting AI retrieval workloads like RAG, semantic search, and hybrid retrieval. It offers filtering, sparse vector support, and flexible deployment via Docker or local builds.

Source: GitHub — github.com/endee-io/endee
1.3k
GitHub stars
1.7k
Forks
C++
Primary language
AGPL-3.0
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Repositoryendee-io/endee
Ownerendee-io
Primary languageC++
LicenseAGPL-3.0 — OSI-approved
Stars1.3k
Forks1.7k
Open issues31
Latest release1.3.5 (2026-05-22)
Last updated2026-06-29
Sourcehttps://github.com/endee-io/endee

What endee is

A C++-based vector database optimized for modern CPU instruction sets (AVX2, AVX512, NEON, SVE2) that implements approximate nearest-neighbor search (ANN/HNSW) with payload filtering and sparse retrieval capabilities. Exposes an HTTP API for index management and vector operations, supports backup workflows, and includes operational logging via MDBX.

Quickstart

Get the endee source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/endee-io/endee.gitcd endee# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

RAG and AI Retrieval Pipelines

Deploy Endee as the retrieval layer for question-answering systems, chatbots, and copilots that need fast vector search with metadata filtering to augment LLM context.

Agentic AI and Agent Memory

Use as persistent long-term memory and context retrieval for autonomous agents built with LangChain, CrewAI, AutoGen, or LlamaIndex to store and rapidly retrieve past observations and tool outputs mid-execution.

Hybrid Search and Semantic Discovery

Combine dense vectors with sparse retrieval and filtering for document search, product discovery, and knowledge base systems where both semantic relevance and exact-term matching improve recall.

Implementation considerations

  • CPU instruction set selection (AVX2, AVX512, NEON, SVE2) must match target deployment hardware; misalignment will degrade performance or cause runtime failures.
  • HTTP API design and payload filtering semantics require upfront schema planning and client-side query translation; no ORM or query builder mentioned.
  • Backup and restoration workflows are documented but require manual orchestration; assess operational readiness before production deployment.
  • Project note indicates active codebase has diverged from this repository; verify feature parity and stability status before adopting for new projects.
  • Dense + sparse vector support requires dual-pipeline indexing and careful relevance tuning; hybrid search relevance requires empirical validation for your domain.

When to avoid it — and what to weigh

  • Proprietary/Commercial Derivative Without Disclosure — AGPLv3 requires source code disclosure if you modify and run Endee as a network service. Evaluate license obligations before embedding in a closed-source managed offering.
  • Need Permissive Licensed Dependency — If your stack requires MIT, Apache 2.0, or BSD licensing only, AGPLv3's copyleft and network-use clause may conflict with your compliance framework; requires legal review.
  • Sub-Million Vector Workloads with Limited Operations Staff — Endee is optimized for billion-scale vectors. For simpler use cases requiring minimal operational overhead, lighter alternatives (e.g., Pinecone, Weaviate in managed form) may reduce deployment friction.
  • Immediate Enterprise Support and SLA Guarantees — Project is community-driven with no stated commercial support tier in the data provided. Critical production environments may require vendor backing; contact [email protected] for terms.

License & commercial use

Licensed under GNU Affero General Public License v3.0 (AGPLv3). This is a copyleft license requiring source disclosure of any modifications if the software is run as a network service. The project explicitly states that trademark/branding permissions are separate; contact [email protected] for clarification.

AGPLv3 is not a permissive OSI license suitable for unmodified commercial closed-source products. Internal use is permitted; network service deployments must disclose modifications. Selling a managed/hosted version requires either full source disclosure to users or a separate commercial license agreement. No commercial support tier, SLA, or dual-licensing arrangement is mentioned in the provided data. Legal and commercial review is mandatory before committing to a revenue model.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityNeeds review
Deployment complexityModerate
DEV.co fitGood
Assessment confidenceHigh
Security considerations

No explicit security audit, vulnerability disclosure policy, or threat model provided in the data. AGPLv3 source availability aids community review. Authentication mechanisms are mentioned but not detailed. For sensitive AI workloads, assess data encryption at rest/in transit, access control granularity, and audit logging capabilities independently. No claims of HIPAA, SOC 2, or other compliance certifications are made.

Alternatives to consider

Pinecone

Fully managed vector database with SLA-backed support, native multi-tenancy, and serverless scaling. No deployment overhead but proprietary and requires commercial commitment.

Weaviate

Open-source vector database (BSL/AGPL hybrid) with multi-vectorization support, GraphQL API, and stronger Kubernetes-native design. More mature enterprise tooling.

Milvus

Open-source (AGPL) vector database with distributed architecture, Kubernetes operator, and broader cloud vendor integrations. Better suited for multi-node high-availability setups.

Software development agency

Build on endee with DEV.co software developers

Start with the quick-start guide, review AGPLv3 license obligations, and assess single-node capacity and sparse/hybrid search requirements against your use case. Contact the Endee team at [email protected] for commercial licensing and support.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

endee FAQ

Can I use Endee in a proprietary SaaS product?
Not without legal review. AGPLv3 requires source code disclosure if you modify Endee and run it as a network service. A commercial license or strict internal-use-only deployment may apply; contact [email protected] for commercial terms.
What is the maximum vector count and latency?
Designed to handle up to 1B vectors on a single node; specific p99 latency and throughput benchmarks are not provided in the README. Benchmarks are marked "Coming Soon" on the website.
Does Endee support clustering or replication?
Not clearly stated in the provided data. The documentation focuses on single-node deployment and backup workflows; multi-node clustering and high-availability setup are not mentioned. Requires documentation review.
How does the sparse vector / hybrid search feature work?
Sparse retrieval support is documented in docs/sparse.md and combined with dense vector search for hybrid results. Specific hybrid ranking strategy and relevance tuning guidance are not summarized in the README; full docs required.

Custom software development services

Need help beyond evaluating endee? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and vector databases integrations — and maintain them long-term.

Ready to Evaluate Endee for Your AI Retrieval Workload?

Start with the quick-start guide, review AGPLv3 license obligations, and assess single-node capacity and sparse/hybrid search requirements against your use case. Contact the Endee team at [email protected] for commercial licensing and support.