Qdrant
ActiveDescription
Qdrant is a high-performance, massive-scale vector database and vector search engine written in Rust, built for the next generation of AI applications.
Key Features
- High-performance ANN — HNSW and scalar quantization for millisecond vector search at scale
- Rich filtering — payload filters, geo queries and hybrid sparse-dense retrieval
- Multimodal & multi-vector — sparse vectors, named vectors and recommendation-style multi-vector recall
- Rust core — memory-safe, zero-cost abstractions, efficient CPU/GPU utilisation
- Cloud-native — sharding, replication, snapshots, distributed deployment and horizontal scaling
- Ecosystem — Python/JS/Rust/Go SDKs plus Qdrant Cloud and Hybrid Cloud offerings
Use Cases
💡 Powering RAG applications with billion-scale vector retrieval and filtering
💡 Building vector-based similarity search for products, documents and images
💡 Combining collaborative filtering and multimodal recall for recommendation systems
💡 Self-hosting a vector database in private or edge environments
💡 Combining sparse retrieval with dense vectors for hybrid search systems
Categories
Quick Start
# Start Qdrant with Docker
docker run -p 6333:6333 -p 6334:6334 \
-v $(pwd)/qdrant_storage:/qdrant/storage \
qdrant/qdrant
# Python SDK example
pip install qdrant-client
from qdrant_client import QdrantClient
from qdrant_client.models import PointStruct, VectorParams, Distance
client = QdrantClient('localhost', port=6333)
client.create_collection('docs', vectors_config=VectorParams(size=384, distance=Distance.COSINE))
client.upsert(
collection_name='docs',
points=[PointStruct(id=1, vector=[0.1]*384, payload={'title': 'Hello'})],
)
hits = client.search(collection_name='docs', query_vector=[0.1]*384, limit=5)