AI agents for Upstash Vector

Get index info

Returns the index's vectorCount, pendingVectorCount (still being indexed), indexSize in bytes, dimension, similarityFunction, indexType (DENSE, SPARSE or HYBRID), the embedding model when it has one.

Search by vector

Finds the vectors most similar to a query vector (numbers, with the index's dimension), as result (most similar first).

Search by text

Searches with plain text on an index created with an embedding model (Upstash embeds the text), as result (most similar first).

Get vectors by ID

Reads vectors by ids, or up to 1,000 whose IDs start with prefix, as result.

Browse vectors

Pages through the vectors of a namespace, as result.vectors (each with id and, when asked, metadata, data and vector) and result.nextCursor.

Add or replace vectors

Writes up to 1,000 vectors, each with an id, a vector (numbers, with the index's dimension), optional metadata (an object to filter on) and optional data (text kept with the vector).

Add or replace text

Writes up to 1,000 texts to an index created with an embedding model (Upstash embeds each text and keeps it as the vector's data), each with an id, the text and optional metadata.

Delete vectorsDeletes data

Permanently deletes vectors from one namespace by ids, by an ID prefix, or by a metadata filter like status = 'archived' (a full scan, slow on large indexes).