The AI-Native Search Database
Unifies relational, vector, text, JSON and GIS in a single engine, enabling hybrid search and in-database AI workflows.
pip install pyseekdb
GitHubBuild Fast
From prototype to production in minutes: create AI apps using Python, run VectorDBBench on 1C2G.
Hybrid Search
Combine vector search, full-text search and relational query in a single statement.
Multi-Model
Support relational, vector, text, JSON and GIS in a single engine.
AI Inside
Run embedding, reranking, LLM inference and prompt management inside the database, supporting a complete document-in/data-out RAG workflow.
SQL Inside
Powered by the proven OceanBase engine, delivering real-time writes and queries with full ACID compliance, and seamless MySQL ecosystem compatibility.
A minimalist API design that keeps you focused on building your AI
Full-stack AI capabilities for application development—from search to inference
Hybrid Search
- Supports multi-path retrieval in a single SQL query, combining vector-based semantic search with keyword-based search for optimized recall.
- Query reranking supports weighted scores, Reciprocal Rank Fusion (RRF), and LLM-based reranking for enhanced results.
- Relational filters are pushed down to storage for optimized performance, and multi-table joins allow relational data retrieval.
Vector & Full-Text Search
- Supports dense vectors and sparse vectors, with multiple distance metrics including Manhattan, Euclidean, inner product, and cosine similarity.
- Vector indexes support in-memory types such as HNSW, HNSW-SQ, HNSW-BQ, and disk-based types including IVF and IVF-PQ, optimizing storage costs.
- Full-text search supports keyword, phrase, and Boolean queries, with BM25 ranking for relevance.
AI Functions
- Manage built-in AI services via the DBMS_AI_SERVICE package in SQL, and register external LLM services.
- Convert text to vector embeddings directly in SQL using the AI_EMBED function.
- Generate text in SQL with AI_COMPLETE, supporting reusable prompt templates.
- Rerank text using LLM-based models in SQL via AI_RERANK.
Semantic Search Engine
Traditional keyword search struggles to capture intent. Semantic search leverages embeddings and vector search to understand meaning and connect text, images, and other modalities. seekdb's hybrid search and multi-model querying deliver more precise, context-aware results across complex search scenarios.
Flexible deployment modes that support everything from rapid prototyping to large-scale user workloads, meeting the full range of your application needs
Embedded Mode
seekdb embeds as a lightweight library installable with a single pip command, ideal for personal learning or prototyping, and can easily run on various end devices.
Client/Server Mode
A lightweight and easy-to-use deployment mode recommended for both testing and production, delivering stable and efficient service.
Seamless Migration to OceanBase Distributed Database
An enterprise-grade, scalable solution that enables smooth application migration to the OceanBase distributed database and handles massive user traffic with ease.
