infiniflow/infinity

The AI-native database built for LLM applications, providing incredibly fast hybrid search of dense vector, sparse vector, tensor (multi-vector), and full-text.

What it solves

Infinity addresses the need for high-performance data retrieval in LLM applications, specifically for Retrieval-augmented Generation (RAG). It solves the challenge of managing and searching across diverse data types—such as dense and sparse embeddings, tensors, and full-text—within a single, fast database system.

How it works

It functions as an AI-native database that supports hybrid search, allowing users to combine different search modalities (dense, sparse, tensor, and full-text) with filtering. It uses a single-binary architecture for easy deployment and provides an intuitive Python API, allowing it to be embedded as a module or run as a separate server via Docker.

Who it’s for

This tool is designed for AI developers building LLM-powered applications such as conversational AI, copilots, recommenders, question-answering systems, and content generation tools.

Highlights

  • High Performance: Achieves 0.1ms query latency on million-scale vector datasets and 1ms latency for full-text search on 33M documents.
  • Hybrid Search: Supports combining dense embeddings, sparse embeddings, tensors, and full-text search.
  • Advanced Reranking: Includes support for rerankers like ColBERT, RRF, and weighted sum.
  • Developer Friendly: Offers a single-binary architecture with no dependencies and a dedicated Python SDK.

Related

  • Project
    endee-io/endee

    A high-performance AI search and intelligence platform that provides dense, sparse, and hybrid vector retrieval for RAG pipelines and AI agent memory.

  • Project
    neuml/txtai

    An all-in-one AI framework for semantic search, LLM orchestration, and language model workflows, featuring a built-in embeddings database for RAG and autonomous agents.

  • Project
    datalevin/datalevin

    Datalevin is a fast, durable Datalog database that provides a declarative alternative to SQL and graph databases, featuring built-in vector search and AI-native capabilities.

  • Project
    meilisearch/meilisearch

    A lightning-fast search engine that provides full-text, semantic, and hybrid search capabilities with search-as-you-type and typo tolerance for applications.

  • Project
    alibaba/zvec

    Zvec is an open‑source, in‑process vector database from Alibaba. It stores dense or sparse embeddings, offers lightning‑fast ANN and full‑text search, supports hybrid queries, and provides SDKs for Python, Node.js, Go, Rust, and Dart. The library runs locally on Linux, macOS, Windows, Android, and iOS, with durable WAL storage and concurrent read access. Ideal for LLM‑based retrieval, edge devices, or any app needing fast, embedded similarity search.