Payload-aware filtering
Store JSON payloads next to vectors and filter on structured metadata such as keywords, numbers, geo values, datetimes, and UUIDs.
Qdrant
Freemium
Qdrant é um mecanismo de busca vetorial open source escrito em Rust para recuperação com AI, busca semântica e correspondência por similaridade. Combina busca vetorial com filtragem de payloads JSON e oferece implantações em nuvem, híbridas, privadas e edge.
O que é Qdrant?
Qdrant is an open-source vector search engine written in Rust. It is designed for vector similarity search at scale and is positioned for AI retrieval workflows that need fast search, metadata filtering, and flexible deployment.
The product combines vector storage with JSON payloads, hybrid dense-plus-sparse search, and multivector retrieval. It also supports cloud, hybrid, private, and edge deployment models, so teams can use it as a managed service or run it in their own infrastructure.
O que Qdrant pode fazer?
Hybrid search
Combine dense and sparse retrieval in a single query for keyword-plus-semantic search workflows.
Multivector retrieval
Use multiple vectors per object to model richer representations and late interaction retrieval patterns.
One-stage filtering
Apply filters during HNSW traversal to support efficient one-stage filtering with high recall and low latency.
Reranking and result control
Re-rank and diversify results with business logic, late interaction models, and techniques such as MMR.
Developer APIs and tools
Work through REST, gRPC, and official clients, with a built-in web UI for inspecting collections and queries.
Casos de uso
“RAG and GenAI”
Use Qdrant as the retrieval layer for RAG systems that need fresh context, metadata filters, and a mix of dense and sparse signals.
“AI agents”
Build memory-backed agents that search for relevant context quickly across evolving conversation or task data.
“Semantic search”
Power semantic search experiences that go beyond keyword matching by combining vector search with metadata constraints.
“Recommendations”
Support recommendation flows where similarity search is used to surface related items and personalized results.
“Anomaly detection”
Analyze records for unusual patterns by searching for points that differ from the expected vector distribution.
Perguntas frequentes
What pricing and deployment options does Qdrant offer?
Qdrant Cloud offers a free tier for testing and prototypes, a standard tier for production workloads, and premium and enterprise options for additional security and compliance needs. The pricing page also describes Hybrid Cloud and Private Cloud deployment options.
Can existing OSS deployments move to Qdrant Cloud?
Yes. The pricing page says you can migrate from OSS to Qdrant Cloud, and it mentions a migration tool and documentation to help with the transition.
What kinds of metadata can Qdrant store and search on?
Qdrant stores payloads as JSON alongside vectors. Those fields can be used for filtering and ranking, and the documentation shows support for types such as integers, floats, booleans, keywords, geo values, datetimes, and UUIDs.
Does Qdrant support hybrid search?
Qdrant supports hybrid queries that combine dense and sparse vectors in one query, including approaches such as BM25, SPLADE++, and miniCOIL.
Where can Qdrant be deployed?
Qdrant can be deployed as managed cloud, hybrid cloud, private cloud, or edge, depending on the level of control and infrastructure model you need.
Informações rápidas
- Category
- Vector Search Engine
- Platform
- Cloud, hybrid, private cloud, and edge
- Primary users
- Developers and teams building AI retrieval systems
- Primary language
- Rust
- Source domain
- qdrant.tech
- Pricing model
- Free tier plus usage-based and enterprise options
Análise de tráfego de Qdrant
Os dados de tráfego são apenas para referência.
- Classificação do domínio
- 79
Alternativas ao Qdrant
博查AI搜索
bochaai.com
博查AI搜索 é um mecanismo de busca multimodal para aplicações de IA que conecta desenvolvedores e agentes a conhecimento externo e resultados de busca.
cito
cito.fim.ai
Cito é um serviço de busca programática de artigos acadêmicos que combina pesquisa semântica e por palavras-chave em um corpus local com cerca de 148 milhões de artigos. Pesquisadores e desenvolvedores podem encontrar artigos por significado ou termos exatos, com metadados e links das fontes em cada resultado.
Waldo
waldo.fyi
Plataforma agêntica de inteligência de marca para pesquisa, propostas, monitorização e projetos personalizados.
Zocket
zocket.com
Zocket é um sistema operacional de marketing com IA para marcas enterprise, reunindo inteligência de marca, produção criativa, pesquisa de consumidores e mídia paga em uma única plataforma. Também oferece planos de preços publicados e uma opção Enterprise para contato com vendas.
Andri
andri.ai
Andri é uma plataforma de IA jurídica agêntica para profissionais do direito no Reino Unido e na Europa. Oferece pesquisa jurídica, elaboração e revisão de documentos e simulação de casos, com teste gratuito sem cartão de crédito.
Linkup
linkup.so
API de busca na web para IA com respostas fundamentadas e trechos de texto completo