Capability · Framework — rag

Pinecone

Pinecone is a fully managed, cloud-native vector database designed for AI workloads. Its serverless tier decouples storage and compute so you pay for what you query, scaling to billions of vectors with low-latency filtering and hybrid sparse-dense search. It supports namespaces for multi-tenancy, integrated inference, knowledge bases, and ships SDKs for Python, Node.js, Java, Go, and REST.

Framework facts

Category: rag
Language: Python / TypeScript
License: commercial
Repository: https://github.com/pinecone-io/pinecone-python-client

Install

pip install pinecone
# or
npm install @pinecone-database/pinecone

Quickstart

from pinecone import Pinecone, ServerlessSpec

pc = Pinecone(api_key='...')
pc.create_index(
    name='docs',
    dimension=1536,
    metric='cosine',
    spec=ServerlessSpec(cloud='aws', region='us-east-1')
)
index = pc.Index('docs')
index.upsert([('id1', [0.1]*1536, {'text': 'hello'})])
res = index.query(vector=[0.1]*1536, top_k=3)

Alternatives

Qdrant — open-source, self-hostable
Weaviate — open-source with hybrid search
Chroma — embedded / open-source
pgvector — Postgres extension

Frequently asked questions

Is Pinecone open source?

No. Pinecone is a commercial managed service. If you need open source or self-hosting, consider Qdrant, Weaviate, or Chroma. The tradeoff is Pinecone's serverless elasticity and zero-ops.

Does Pinecone do hybrid search?

Yes. Pinecone supports sparse-dense hybrid search — you index both dense embeddings and sparse (BM25-style) vectors and can weight their contribution at query time. Useful for RAG where keyword precision matters.

Sources

Pinecone — docs — accessed 2026-04-20
Pinecone — site — accessed 2026-04-20