Capability · Framework — rag

Qdrant

Qdrant is one of the fastest open-source vector databases, written in Rust with a focus on filtered search and production reliability. It supports dense, sparse, and multi-vector collections, hybrid search with fusion, scalar/product/binary quantisation, snapshots, distributed deployments, and rich JSON payload filtering with geo/temporal predicates. Qdrant Cloud and Qdrant Hybrid Cloud provide managed deployments.

Framework facts

Category: rag
Language: Rust (Python / TS clients)
License: Apache 2.0
Repository: https://github.com/qdrant/qdrant

Install

pip install qdrant-client
# run server:
# docker run -p 6333:6333 qdrant/qdrant

Quickstart

from qdrant_client import QdrantClient
from qdrant_client.models import VectorParams, Distance, PointStruct

client = QdrantClient(url='http://localhost:6333')
client.recreate_collection(
    collection_name='docs',
    vectors_config=VectorParams(size=1536, distance=Distance.COSINE)
)
client.upsert('docs', points=[PointStruct(id=1, vector=[0.1]*1536, payload={'text': 'hi'})])

Alternatives

Pinecone — managed alternative
Weaviate — open-source with hybrid + modules
Milvus — open-source at huge scale
Chroma — embedded alternative

Frequently asked questions

Qdrant or Weaviate?

Both are great open-source options. Qdrant tends to win on raw QPS and filtered-search latency, especially with heavy payload filters. Weaviate has richer built-in modules (hybrid BM25+vector, named vectors, generative search) and a more batteries-included feel.

Does Qdrant support multi-tenancy?

Yes — via payload-based tenant IDs with fast filtered search, and via collection-per-tenant setups for stricter isolation. Qdrant Cloud and Hybrid Cloud add authentication, role-based access, and private networking.

Sources

Qdrant — docs — accessed 2026-04-20
Qdrant — GitHub — accessed 2026-04-20