PineconeBuilding block

Pinecone is a managed vector database for semantic, full-text (BM25) and hybrid search. It offers hosted embedding and reranking models, namespaces and metadata filtering, with Python and JavaScript SDKs and a REST API.

Used in AI recipes forEmbeddings Search

Website Docs Changelog Compare with others

Pinecone at a glance

From official pages, checked Sep 29, 2026. Tap the arrow for each source.

Free tier: Yes
Starter plan: up to 2 GB storage, 2M write units and 1M read units per month, 5 indexes.
Starting price: $20/month
Builder plan, flat monthly price.
Rate limit: 100 queries/sec per namespace
Starter plan; upserts also 100 requests/sec per namespace.
Size limit: top_k max 10,000
Dense vectors up to 20,000 dimensions; 40 KB filterable metadata per record.
Storage: 2 GB free
Starter plan storage cap.
Official SDKs: PythonJavaScript / TypeScript
Trains on your data: No
Security page says customer data is only used for servicing API calls.
EU hosting: Yes
eu-west-1 (Ireland) and europe-west4 (Netherlands) on Builder plan and above.
Compliance: SOC 2ISO 27001GDPRHIPAA

What you can build with Pinecone

PineconeRecipeBuild

Semantic search over your own text with Pinecone

Chunk text, embed it with Pinecone's hosted llama-text-embed-v2 model, and query by meaning. No separate embedding provider.

Free tier: YesFrom $20/month

EmbeddingsSearch

Pinecone at a glance

What you can build with Pinecone

Semantic search over your own text with Pinecone

Update timeline