⚡Stack Radar

Infrastructure & Security · Databases & Storage

Vector Databases

10 paid against 11 free you can run yourself.

Paid against free › Infrastructure & Security › Databases & Storage › Vector Databases · Data and databases on Repo Radar

Vector databases store embeddings and find the most similar items fast, which powers semantic search, recommendations and retrieval for AI. The paid options are managed services from the main open-source engines (Qdrant, Weaviate, Zilliz for Milvus, Chroma, Vespa), serverless specialists (Pinecone, turbopuffer, Upstash) and databases that add vector search to what you already run (MongoDB Atlas, Tiger Cloud for Postgres). On the open-source side you can self-host Qdrant, Milvus, Weaviate, Chroma or Vespa, add vectors to Postgres with pgvector and pgvectorscale, embed search in your app with LanceDB or sqlite-vec, or build on the FAISS library. Marks come from vendor pricing pages and project READMEs as of 19 October 2026; nothing was tested hands-on.

Paid 10 reviewed

Prices and free tiers as checked, dated on each review.

PPinecone

Managed vector database and AI knowledge platform for production AI applications.

$50free tierQQdrant

AI-native vector search engine for semantic search and retrieval.

free tierWWeaviate

Open-source AI database for vector search, hybrid search, and AI application development.

$45free tierZZilliz Cloud

Fully managed vector database for AI applications

$197free tierTturbopuffer

Fast, object-storage native search engine for AI and large-scale data.

—CChroma

Open-source search infrastructure for AI, supporting vector, full-text, and metadata search.

$250free tierUUpstash Vector

Serverless vector database with per-request pricing for AI applications.

$60free tierMMongoDB Atlas Vector Search

Vector search integrated with MongoDB Atlas for AI applications.

$30free tierTTiger Data

Managed and self-hosted PostgreSQL for time-series, events, and analytics.

$30free tierVVespa

Vespa is an AI search platform for retrieval, ranking, and inference.

free tier

Free 11 direct · 15 related

Free to use. You run it on your own server.

Direct alternatives

They do this job.

FAfacebookresearch/faiss

A library for efficient similarity search and clustering of dense vectors.

★ 41.2kFreeQDqdrant/qdrant

It stores lists of numbers that stand for meaning, and finds the closest matches.

★ 35.1kFreePGpgvector/pgvector

A Postgres extension that adds a vector column type and nearest-neighbour search.

★ 23.3kFreeLAlancedb/lancedb

An embedded database for embeddings, imported as a library rather than run as a server.

★ 11.6kFreeFAFalkorDB/FalkorDB

A graph database that keeps the connections as sparse matrices and answers queries with linear algebra.

★ 10.1kFreeSQasg017/sqlite-vec

Adds find-the-closest-match to SQLite as one small portable file, with no server and nothing to compile.

★ 8.2kFreeVEvespa-engine/vespa

The AI search platform

★ 7.1kFreeUSunum-cloud/usearch

A small, fast engine for finding the closest matches among millions of number lists.

★ 4.3kFreePGtimescale/pgvectorscale

A PostgreSQL add-on that makes similarity search work on more data than fits in memory.

★ 3.1kFreeVEsupervc-stack/VectorChord

A PostgreSQL index for similarity search that lives on disk rather than in memory.

★ 1.8kFreeSQsqliteai/sqlite-vector

An add-on that gives the single-file SQLite database meaning-based search, with its own accuracy figures printed in the README.

★ 1.1kFree

Related alternatives

A second choice: they cover part of this job or are a building block for it.

MEmeilisearch/meilisearch

A search engine you run yourself, built for fast answers and forgiving spelling.

★ 59.2kFreeHIvectorize-io/hindsight

A memory store for AI agents that keeps what matters across sessions instead of replaying the whole history.

★ 49.4kFreeDUduckdb/duckdb

An analytical SQL database that runs inside your own process and queries files directly.

★ 42.1kFreeTIpingcap/tidb

TiDB is built for agentic workloads that grow unpredictably, with ACID guarantees and native support for transactions, analytics, and vector search. No data silos. No noisy neighbors. No infrastructure ceiling.

★ 40.7kFreeLIHKUDS/LightRAG

A retrieval system that builds a network of things and relationships out of documents, then answers questions against it.

★ 40.1kFreeGRmicrosoft/graphrag

A pipeline that builds a knowledge graph from documents and answers questions from nested cluster summaries.

★ 36.3kFreeTUtursodatabase/turso

A rewrite of SQLite in Rust that reads the same files and has not yet reached version 1.0.

★ 24.7kFreeTItimescale/timescaledb

A PostgreSQL add-on that makes years of time-stamped rows fast to query.

★ 23.7kFreeBEbeekeeper-studio/beekeeper-studio

A desktop window onto a database, so you can read and edit rows without writing commands.

★ 23.6kFreePAparadedb/paradedb

A PostgreSQL extension that adds search-engine-quality keyword search as ordinary SQL.

★ 9.4kFreeEMapple/embedding-atlas

Apple's interactive embedding map with automatic cluster labelling — worth studying as a design reference for an overview screen rather than adding as a dependency.

★ 5kFreeCHfeyninc/chonkie

Focused text-chunking library — semantic and recursive splitters that respect message boundaries, so subtopics come out clean.

★ 4.7kFreeCRbmoscon/cryptofeed

Receives live market data from 28 exchanges, normalises it to one shape, and can write it straight into nine kinds of store.

★ 2.9kFreeMOMinishLab/model2vec

Distils a sentence model into a static lookup table, so producing embeddings needs no graphics card.

★ 2.2kFreeVIMinishLab/vicinity

A small Python library giving one interface over several nearest-neighbour search methods.

★ 351Free

Feature by feature

Every paid tool and free repository we compared, marked against the same features.

✓has itPartlysome plans or extra work—confirmed not available?not confirmed✓the vendor's claim onlypricingClick a mark for its note and source

Paid · 10Free · 10
Featureqdrant/qdrantmilvus-io/milvusweaviate/weaviatechroma-core/chromapgvector/pgvectorlancedb/lancedbtimescale/pgvectorscaleasg017/sqlite-vecvespa-engine/vespafacebookresearch/faiss
Starts at$20 per month flat (Builder)Standard tier, usage-based$45 per month (Flex)Serverless from $0 per month (usage)$16 per month minimum usage (Launch)$0 per month plus usage (Starter)$0.4 per 100K requests (Pay as You Go)Flex: $0.011 per hour, up to $30 per monthCompute from $30 per month plus storageFrom $0.05 per vCPU-hour (Startup)Apache-2.0Apache-2.0BSD-3-ClauseApache-2.0PostgreSQL LicenseApache-2.0PostgreSQL LicenseApache-2.0Apache-2.0MIT
Deployment
Managed cloud serviceThe vendor runs the database for you.
Pricing
Free tierA free plan or free database to start.
Serverless usage pricingPay for reads, writes, storage or requests rather than fixed servers.
Deployment
Dedicated clustersReserved, isolated resources for predictable performance.
Bring your own cloudThe managed service runs inside your own cloud account.
Search
Hybrid searchCombines vector search with keyword or full-text search.
Metadata filteringFilter vector results by attributes.
Built-in embeddingsCreates embeddings for you, without a separate model service.
Scale
Multi-tenancyIsolates many tenants or namespaces in one database.
Disk or tiered storageKeeps large indexes on disk or object storage to cut memory cost.
Quantisation or compressionShrinks vectors to save memory and cost.
Integration
Postgres or SQL integrationVector search inside Postgres, SQLite or with SQL.
Deployment
Embedded or in-processRuns inside your application without a separate server.
Self-hostingRun it on your own machines.
Security
Enterprise security (SSO, HIPAA)Single sign-on, audit logs or HIPAA eligibility.
Licence
Open sourceSource code published under an open licence.

Excluded candidates

Looked at, and left out of the comparison, with the reason.

Paid apps

  • LanceDB CloudLanceDB's pricing address now redirects to a contact form, so its managed service has no published plans to compare; the open-source LanceDB is selected as G6.Source
  • MarqoMarqo's pricing address redirects to a demo booking page, and the open-source Marqo now describes itself as ecommerce search and discovery rather than a general vector database.Source
  • DataStax Astra DBDataStax's pricing address now redirects to IBM watsonx.data pricing after IBM's acquisition, so Astra DB has no standalone vector database plan to compare.Source

Repositories

  • unum-cloud/usearchUSearch is a fast single-file search and clustering engine for vectors. It is an index library like FAISS (G10), which is more widely used, so it ranked below it.Source
  • supervc-stack/VectorChordVectorChord is a disk-friendly Postgres vector extension under a dual licence. It overlaps pgvector and pgvectorscale, which are more widely used, and has had no push since 6 August 2026.Source
  • FalkorDB/FalkorDBFalkorDB is a graph database built on GraphBLAS. Vector search is a side feature of a graph database, which fits graph databases better than this comparison.Source
  • sqliteai/sqlite-vectorSQLite-Vector is a SQLite vector extension like sqlite-vec (G8), with a smaller community; sqlite-vec represents the approach here.Source
  • alibaba/zveczvec is a lightweight in-process vector database from Alibaba. It overlaps LanceDB (G6) as an embedded option and is younger, so it ranked below it.Source
  • meilisearch/meilisearchMeilisearch is a search engine API that adds AI-powered hybrid search to keyword search. It is primarily a site and app search engine, compared better with search tools.Source
  • typesense/typesenseTypesense is an open-source search engine positioned against Algolia and Pinecone. Vector search is one feature of a keyword search engine, which fits search tools better.Source
  • redis/redisRedis is an in-memory data store with vector search among many features. It is a general cache and database rather than a vector database first.Source
  • valkey-io/valkey-searchvalkey-search adds vector search to the Valkey key-value store. It is a module for one data store with about 150 stars, too narrow for this comparison.Source
  • RyanCodrai/turbovecturbovec is a vector index built on TurboQuant, written in Rust. It is an index library like FAISS rather than a database, and young.Source
  • volcengine/OpenVikingOpenViking is a context database for AI agents that unifies agent memory and resources. It is agent memory infrastructure rather than a general vector database.Source

Gaps

What the research could not settle from the vendors' own pages.

What vendors don't publish

  • Vector databases bill in different units (read and write units, GiB written, requests, compute units, GB of dedicated capacity, vCPU-hours), so the same workload cannot be priced across vendors without each calculator.Pinecone, Weaviate, Zilliz Cloud, Chroma, Upstash Vector, Vespa · checked 2026-10-19 · source 1 · source 2 · source 3 · source 4
  • Qdrant Cloud's Standard and Premium tiers show no fixed price on the pricing page; the cost comes only from its calculator, and Premium requires an unstated minimum spend.Qdrant · checked 2026-10-19 · source 1

Paid against free, in your inbox

One job at a time: the paid tools worth paying for, the free ones you can run yourself, and who should pick which.

No tracking pixels. One click to leave. We use your address to send Stack Radar and nothing else. Privacy.