Manticore Search

Active
GitHub C++ GPL-3.0

Description

Manticore Search is an open-source lightweight search database with native full-text, vector and hybrid retrieval, commonly used as agent/RAG backend.

Key Features

  • Full-text + vector + hybrid — unified interface for all three
  • MySQL-compatible protocol — reuse MySQL clients and tools
  • High-performance — C++ core, millisecond latency
  • Real-time indexing — inserts and updates visible within milliseconds
  • Horizontal scale — sharding and replicas for large workloads
  • Multilingual — Chinese and English tokenization and synonym support

Use Cases

💡 Backing RAG with full-text + vector hybrid retrieval
💡 A lightweight replacement for ElasticSearch
💡 Migrating ecommerce or media search onto Manticore
💡 A low-latency retrieval API for agents

Strengths & Limitations

Strengths

  • Actively maintained, recent updates
  • High community interest (11.9k stars)

⚠️ Limitations

  • High issue backlog (707 open issues)
  • Restrictive license (GPL-3.0)

Quick Start

docker run -d --name manticore -p 9306:9306 -p 9308:9308 manticoresoftware/manticoresearch:latest
mysql -h0 -P9306 -e "CREATE TABLE products(title text, vec float_vector);
  INSERT INTO products(title,vec) VALUES ('hello',(0.1,0.2));"

Related Projects

Meilisearch

59.0k · Rust
Active A

Meilisearch is a lightweight open-source search engine with typo tolerance, native vector search and instant indexing, commonly used as the retrieval layer for agents and RAG.

search-enginefull-text-searchvector-search +2
  • · Typo-tolerant — forgiving search out of the box
  • · Instant indexing — new or updated docs are searchable within milliseconds
  • · Native vector search — built-in ANN vector index, hybrid with full-text