Vectors

Vector types support embeddings, metadata, quantization, and similarity search.

See Type construction and operations for metric definitions, dimensions, exact-scan behavior, HNSW graph construction/search, quantization, result ordering, and the current protocol’s lack of metadata-filter expressions.

Type	Construction	Principal operations
dense_vector	SET	DISTANCE
sparse_vector	SET	GET
vector_set	CREATE	ADD, REMOVE, SEARCH, LEN
vector_metadata	CREATE	GET, SET, DELETE
quantized_vector	SET	DEQUANTIZE