pyobvector

What it is and what it does

pyobvector is a Python client library for OceanBase's vector storage capabilities, enabling approximate nearest neighbor (ANN) search, full-text search, and JSON queries on OceanBase databases. It wraps OceanBase's vector indexing (HNSW, VSAG) and exposes it through two main interfaces: a Milvus-compatible API for drop-in compatibility with existing Milvus code, and a SQLAlchemy hybrid mode that lets you mix vector operations with standard SQL queries on the same tables.

The package is built on top of SQLAlchemy, aiomysql, and pymysql for database connectivity, and numpy for vector operations. It supports four modes of use: Milvus-compatible collections, SQLAlchemy ORM with vector extensions, embedded local SeekDB (no server), and hybrid search combining full-text and vector similarity. Installation is straightforward (pure Python wheel), and it requires Python 3.10+.

Use it for:

• Migrate from Milvus to OceanBase without rewriting client code by using MilvusLikeClient.
• Build semantic search on relational data by combining vector similarity with SQL filters and joins via SQLAlchemy.
• Perform hybrid search combining full-text search and vector similarity in a single query using HybridSearch.
• Prototype vector applications locally with embedded SeekDB before deploying to a server.
• Store and query embeddings alongside structured metadata (JSON fields) in OceanBase tables.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

pyobvector is a Python SDK for OceanBase Vector Store that provides vector similarity search, full-text search, and JSON table operations through SQLAlchemy or a Milvus-compatible API.

Yes, if you are already using OceanBase or planning to adopt it for vector workloads. The Milvus-compatible API lowers switching costs, and SQLAlchemy integration lets you avoid a separate vector database. No security vulnerabilities reported. Active maintenance and low install friction. Not recommended if you need a standalone vector database—pyobvector is a client for OceanBase, not a replacement for it.

Install

pyobvector on PyPI

pip install pyobvector

uv add pyobvector

poetry add pyobvector

Installing pyobvector

Before you install

Low install friction: pure Python wheel with no compiled dependencies. Active maintenance—last commit 2026-07-29, released 16 days ago. Requires Python 3.10+. Six runtime dependencies (aiomysql, numpy, pydantic, pymysql, sqlalchemy, sqlglot) are all stable, widely-used libraries.

License in practice

Licensed under Apache-2.0 (permissive). No restrictions on commercial or private use; modifications and redistribution are allowed under the same license terms.

Quickstart

pip install pyobvector==0.2.29

from pyobvector import MilvusLikeClient

client = MilvusLikeClient(uri="127.0.0.1:2881", user="test@test")
res = client.search(collection_name="test", data=[0,0,0], anns_field='embedding', limit=5)

Requires a running OceanBase server or local SeekDB instance; optional pyseekdb dependency needed for embedded mode (pip install pyobvector[pyseekdb]).

Verify before relying

• Stability and maturity of the OceanBase backend itself—pyobvector is a client SDK only.
• Performance characteristics (latency, throughput) for large-scale vector workloads compared to dedicated vector databases.
• Whether Milvus API compatibility is complete or partial (which Milvus operations are unsupported).

Package facts

Evidence: pyobvector-0.2.29-py3-none-any.whl

Keywords: obvector, oceanbase, vector store

Tags

More Database packages