Qdrant

Qdrant is an open-source vector search engine and vector database for developers, data scientists, and ML engineers building production AI retrieval systems such as semantic search, RAG, and recommendations.

Key capabilities include:

• High-dimensional similarity search using HNSW graphs for sublinear-time retrieval, with support for approximate and exact search and configurable distance metrics (Euclidean, Cosine, Dot Product)
• Memory-efficient indexing with scalar, product, and binary quantization to reduce footprint while maintaining search quality
• Metadata-aware retrieval via JSON payloads, enabling advanced filtering and faceting across boolean, numeric, and geospatial fields
• Real-time operations and scaling, including insert/update/delete without downtime, multi-index collection management, and distributed clustering for horizontal scaling

Pricing: Freemium. Free tier is free forever for testing and prototypes (includes 1GB RAM and 4GB Disk). Paid tiers include Standard (Usage-based pricing) for production workloads and Premium (Minimum spend required) for enterprises needing SSO and private VPC links. On-premise options are also available for self-hosting.

Notable for being written in Rust and designed as a retrieval-first database with flexible APIs (REST and gRPC) and features that support multimodal embeddings (text, images, audio), making it a practical foundation for chatbots, AI agents, anomaly detection, and hybrid search workflows.