Qdrant

hybridFree (self-hosted, Apache 2.0)Free Tier (1GB cluster)Standard Tier (usage-based)Premium Tier (minimum spend required)open source

Best for

High-performance vector search with rich payload filtering — Rust-written, predictable latency, and a clean self-hosted path

Limitations

No built-in embedding generation — bring your own embeddings; managed cloud is younger than Pinecone's; fewer turnkey RAG modules than Weaviate

Features

Hybrid Search
Combined dense vector + sparse/keyword (BM25) search in one query
Metadata Filtering
Pre/post filtering by structured metadata — JSON payloads, tags, dates
Multi Tenancy
Native namespace or tenant isolation for per-user or per-agent vector spaces
Managed Embeddings
Built-in embedding generation at ingest and query time — no separate embedding service required
Mcp Server
Official Model Context Protocol server exposing vector operations as agent tools
Serverless
Pay-per-use scaling with separated storage and compute

Frameworks

langchainllamaindexvercel-aiopenai-agentshaystackdspy

SDK Languages

pythonjavascripttypescriptrustgojavacsharp

Compliance

soc2gdprhipaa

Qdrant

Qdrant is an open-source vector database written in Rust, optimized for predictable latency and rich payload filtering. Its query planner uses payload indexes to short-circuit vector search whenever a metadata filter would dominate — which makes it especially strong for agent workloads where every query is scoped by tenant, user, or document type.

For agents, Qdrant's standout features are dense + sparse + late-interaction hybrid retrieval in a single API, tenant-aware payload filtering, and shard-key-based multi-tenancy that scales horizontally. The official Qdrant MCP server exposes search, upsert, and snapshot operations as agent tools.

Qdrant deliberately stays out of the embedding business — you bring your own embeddings (OpenAI, Cohere, local models). For teams that want the embedding pipeline owned outside the database (versioning, replays, A/B tests), this is a feature, not a missing one. For teams that want one-click ingest of raw text, Weaviate or Pinecone are a closer fit.

Agent-specific features:

  • Hybrid search: dense + sparse + ColBERT-style late interaction in a single query_points call
  • Payload-aware query planner: heavy filters short-circuit the vector search
  • Shard-key multi-tenancy with per-tenant snapshots and deletion
  • Official MCP server (mcp-server-qdrant) for agent-driven retrieval
  • Hybrid Cloud and Private Cloud for fully managed deployments inside your VPC
Last verified: 2026-06-10Verified by: editorial