Module index 

Vector index management for pgvector.

pgvector supports two approximate nearest-neighbor (ANN) index types:

Index	Algorithm	Best For	Tradeoff
IVFFlat	Inverted file with flat quantization	Large datasets, tunable recall	Requires training data
HNSW	Hierarchical navigable small world	Most workloads, no training needed	Higher memory usage

Choosing an Index

• HNSW is recommended for most use cases — better recall/speed tradeoff, no training step, and supports concurrent inserts.
• IVFFlat is useful when memory is constrained or when you have very large datasets and can tolerate a training step.

Modules

extension

SQL helpers for pgvector extension management.

Structs

BinaryVectorIndex

A binary vector index definition.

BinaryVectorIndexBuilder

Builder for BinaryVectorIndex.

HnswConfig

Configuration for HNSW indexes.

IvfFlatConfig

Configuration for IVFFlat indexes.

VectorIndex

A vector index definition.

VectorIndexBuilder

Builder for VectorIndex.

Enums

IndexType

The type of ANN index to create.