Vector similarity search

Search a collection by vector similarity. Returns the top-k most similar vectors, optionally with metadata.

post/api/v1/query

Request

  • Base URL: https://api.vectorpanda.com
  • URL: https://api.vectorpanda.com/api/v1/query
  • Auth: HTTP bearer

Request body

collectionstring required

Collection name

vectornumber[] required

Query vector (must match collection dimension)

top_kinteger

Maximum number of results to return

similarity_thresholdnumber float

Minimum similarity score to include in results

distance_metric'cosine' | 'euclidean' | 'dot_product'
include_metadataboolean

Include metadata fields in results

use_index'pca' | 'none'

Index strategy override (default chosen automatically)

filterobject

MongoDB-style metadata filter. Only results whose metadata passes all conditions are returned. Operators: $eq, $ne, $gt, $gte, $lt, $lte, $in, $nin. Logical: $and, $or. Shorthand: {"field": "value"} means {"field": {"$eq": "value"}}. Multiple top-level keys are implicitly $and.

Example request

{
  "collection": "my_embeddings",
  "filter": {
    "category": {
      "$eq": "electronics"
    },
    "price": {
      "$lte": 500
    }
  }
}

Response

Search results

Changes

No recorded changes to this operation across all 1 revision of this API.