OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Vectordb — jina-ai

Vectordb is an open-source Python library for storing and searching vectors associated with structured documents. It offers CRUD operations, local and served database workflows, and scaling options including sharding and replication. Its documented search choices include exhaustive search with InMemoryExactNNVectorDB and approximate nearest-neighbour search with HNSWVectorDB.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Vectordb — jina-ai — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Vectordb when an application already has document embeddings and needs a Python-based store and nearest-neighbour retrieval workflow, either locally or through a served database.

Who it is for

  • Python developers building semantic search or embedding-retrieval features.
  • Teams that need a vector index for an AI knowledge or neural-search pipeline.
  • Developers who want to expose vector search to clients through a service or deploy an instance to Jina AI Cloud.

Documented uses

  • Index documents that contain embeddings.
  • Retrieve nearest-neighbour matches for an embedding query.
  • Manage indexed records with create, read, update and delete operations.
  • Serve a vector database over gRPC, HTTP or WebSocket for remote client searches.
  • Configure sharding and replication for a served database.
Before you choose

Requirements & limits

The repository identifies the project as open source under the Apache-2.0 license. The documented Jina AI Cloud deployment path requires a Jina AI Cloud account.

  • The local example uses DocArray schemas and NumPy embedding arrays; embedding generation itself is not covered in the shown workflow.
  • InMemoryExactNNVectorDB performs exhaustive search, while HNSWVectorDB uses approximate nearest-neighbour search and offers tuning settings.
  • The README documents gRPC, HTTP and WebSocket service protocols; its example client uses gRPC.
  • Sharding distributes indexing and search work, while replicas synchronize indexes using RAFT.
  • The README states that Jina AI Cloud deployments use one replica.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Index documents and search locally

Create a local vector database from typed documents and retrieve nearest-neighbour matches for a query embedding.

Before you start

  • Python environment with Vectordb, DocArray and NumPy installed; the README gives pip install vectordb as the installation command.
  • A DocArray document schema containing text and an embedding field, plus document and query embeddings.

Steps

  1. Install Vectordb with the command below.

    Documented command
    pip install vectordb

  2. Define a DocArray BaseDoc schema with a text field and a fixed-size embedding field.

  3. Create an InMemoryExactNNVectorDB or HNSWVectorDB instance using the schema and a workspace path.

  4. Index a DocList of documents, search with a query document, and read matches from the result’s .matches attribute.

Expected outcome

A local index containing the supplied documents and nearest-neighbour matches for a query embedding.

Guide limitations

  • The README’s sample uses randomly generated embeddings, so it demonstrates the indexing and search flow rather than a complete text-embedding pipeline.
  • The sample does not describe how to generate embeddings from application content.
Documentation 1 ↗
Source-based practical guide

Deploy a database instance to Jina AI Cloud

Prepare and deploy a serving Vectordb instance to Jina AI Cloud, then connect a client to its assigned endpoint.

Before you start

  • A Python file containing a Vectordb instance and a serving block protected by an if __name__ == '__main__' guard.
  • A Jina AI Cloud account and the jc command-line tool.

Steps

  1. Add the database instance and serving block to a Python file, following the README’s example.

  2. Sign in with jc login.

  3. Deploy the instance with vectordb deploy --db example:db, substituting the module and instance names used in the file.

  4. Use the assigned deployment ID in a Vectordb Client address to connect.

Expected outcome

A deployed database endpoint that a Vectordb client can address.

Guide limitations

  • The README says Jina AI Cloud deployments use one replica; replication in the cloud is described as not yet enabled in that documentation.
  • This guide covers the documented deployment and connection stages, not account creation details or cloud pricing.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
jina-ai/vectordb · Official homepage title · retained in its source language
How we review sourcesSuggest a correction