OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Vicinity

Vicinity is a lightweight Python vector-store library that provides a shared interface for nearest-neighbor search across multiple backends, with tools to save and load stores and evaluate search speed and recall.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Vicinity — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Vicinity when a Python application needs vector-neighbor search or when you want to compare supported backends through a consistent interface.

Who it is for

  • Python developers adding nearest-neighbor retrieval over vector embeddings.
  • Teams comparing indexing backends or distance metrics for a retrieval component.
  • Researchers measuring backend queries-per-second and recall.

Documented uses

  • Search for the closest stored items using a query vector, top-k count or threshold.
  • Use one interface to experiment with multiple vector-search backends.
  • Evaluate a backend by measuring queries per second and recall.
  • Save and reload vector stores locally or through Hugging Face Hub methods.
Before you choose

Requirements & limits

The project is an open-source Python library under the MIT license, installed with pip.

  • The basic backend uses exact search; HNSW, USEARCH, Annoy, PyNNDescent, FAISS and Voyager provide approximate-search options.
  • Approximate-neighbor backends do not support dynamic deletion; rebuilding the index is required to delete items.
  • Insertion is supported by FAISS, HNSW and USEARCH; the basic backend supports insertion and deletion.
  • Backends and integrations can be installed selectively, while the all extra installs all of them.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install Vicinity and query a vector store

Set up the base Python package and retrieve nearest items for a query vector.

Before you start

  • Python environment with pip
  • Item values and their corresponding vector representations

Steps

  1. Install the base package with the command below.

    Documented command
    pip install vicinity

  2. Import Vicinity, Backend and Metric from vicinity, and provide vectors alongside their items.

  3. Create a store with Vicinity.from_vectors_and_items, choosing a backend and metric.

  4. Call query with a query vector and k for a top-k result, or use query_threshold with a threshold.

Expected outcome

A Vicinity store returns nearest items for a single query vector; its query interface also accepts lists of query vectors.

Guide limitations

  • This guide uses the basic backend from the documented quickstart; other backends may require specific installation extras.
  • The example requires the caller to provide vector representations; the quickstart does not generate embeddings.
Documentation 1 ↗
Source-based practical guide

Measure nearest-neighbor backend performance

Evaluate a Vicinity instance using query throughput and recall.

Before you start

  • Vicinity installed with the selected backend
  • A full vector collection and query vectors

Steps

  1. Create a Vicinity instance using the vectors and selected backend.

  2. Choose query vectors from the collection, as in the documented example using the first 1,000 vectors.

  3. Call vicinity.evaluate with full_vectors and query_vectors.

  4. Read the returned queries-per-second and recall values.

Expected outcome

The evaluation call returns throughput and recall measurements for the selected Vicinity instance.

Guide limitations

  • The README describes the metrics but does not prescribe a benchmark protocol or establish which backend will suit a particular workload.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
MinishLab/vicinity · Official homepage title · retained in its source language
How we review sourcesSuggest a correction