Vector DB AI apps

Fully managed vector database service built on open-source Milvus.

Zilliz Cloud is a fully managed vector database from the creators of Milvus. It runs billion-scale similarity search for RAG, semantic search, and AI applications without the operational burden of self-hosting, offering a free serverless tier, auto-scaling serverless, and dedicated clusters across major clouds. Features include GPU-accelerated indexing, tiered storage, and hybrid dense/sparse search.

Vector + graph database for multimodal AI data.

ApertureDB is a purpose-built database that unifies vector search, a knowledge graph, and multimodal data management — images, video, documents, and their metadata — in a single system. Instead of stitching together a vector store, object storage, and a graph database, teams query embeddings, relationships, and the underlying media together to power GenAI pipelines and agents.

Open-source Postgres vector database for AI apps.

Lantern is a PostgreSQL extension and managed cloud service that turns Postgres into a production vector database for AI. It implements HNSW indexing for fast approximate nearest-neighbour search and layers on embedding generation, vector compression, and hybrid search that combines vector similarity with BM25 full-text and reranking — all inside the database teams already run.

Vector search as a zero-dependency SQLite extension.

sqlite-vec is a small SQLite extension that adds vector search directly to the database engine. Written in pure C with no dependencies, it compiles and runs anywhere SQLite does — Linux, macOS, Windows, the browser via WebAssembly, mobile, and even a Raspberry Pi — and stores float, int8, and binary vectors in ordinary SQLite tables. It ships bindings for Python, Node.js, Ruby, Go, and Rust, making local-first RAG possible without a separate vector database.

Serverless vector database for AI search and RAG.

Upstash Vector is a fully managed, serverless vector database for semantic search, retrieval-augmented generation, and recommendations. It exposes a simple REST API with first-class Python and TypeScript SDKs, stores JSON metadata alongside embeddings for filtered queries, and can generate embeddings for you with built-in models. Billing is pay-as-you-go with a free tier, so there is no infrastructure to provision or scale.

Incremental data framework for fresh AI context.

CocoIndex is an open-source data transformation framework that keeps AI agents and LLM apps supplied with continuously fresh, structured context. It turns sources like codebases, PDFs, databases, and Slack into vector or graph stores, and reprocesses only what changed (delta-only) with parallel execution by default. A Rust core drives reliability while pipelines are defined declaratively in Python, with end-to-end lineage and an observability UI called CocoInsight.

Retrieval engine with hybrid search, multi-vector, and custom ranking in one query.

TopK is a serverless retrieval engine that unifies vector (semantic), keyword (lexical), and multi-vector search with custom ranking in a single query and API — replacing the multi-database stack that RAG and search apps usually stitch together. Storage, inference (embedding, OCR, parsing), and queries run inside the engine or your own VPC, with SDKs for Python, JavaScript/TypeScript, and Rust plus a CLI and an MCP server.

A globally distributed vector database built into Cloudflare Workers.

Vectorize is Cloudflare's vector database for building AI-powered apps on its Workers platform. It stores and queries embeddings for semantic search, recommendation, classification, and RAG, and cross-references results against data in R2, D1, and KV. Embeddings can come from Workers AI or external providers like OpenAI, and indexes are configured via the dashboard, Wrangler CLI, or REST API.

SQL vector database built on ClickHouse — vector, full-text, and analytics in one query.

MyScale is a SQL-native vector database built as a ClickHouse fork, so a single SQL statement can combine vector similarity search, full-text search, and standard structured/analytical queries. It is offered as a managed cloud service with a free starter pod, and its core engine, MyScaleDB, is open source. The SQL-first design lets teams join embeddings with their existing relational data without standing up a separate vector store.

AI-native vector search for multimodal product discovery.

Marqo is an end-to-end vector search engine that generates, stores, and retrieves embeddings for text and image data through a single API, so teams don't have to stitch together a separate embedding pipeline. It has since repositioned around AI-native ecommerce search, merchandising, and recommendations, with turnkey integrations for major commerce platforms.

All-in-one open-source embeddings database for semantic search and RAG.

txtai is an embeddings database that unifies dense and sparse vector indexes, graph networks, and a relational database into a single store for semantic search and LLM workflows. Beyond vector search it ships retrieval-augmented generation, pipelines, and agents, and can be built with Python or declarative YAML. It runs locally or self-hosted and exposes API bindings for JavaScript, Java, Rust, and Go.

Multimodal database for AI — vectors plus raw data, versioned.

Deep Lake, by Activeloop, is a database for AI that stores vectors alongside raw multimodal data — text, image, video, audio, and metadata — in a single version-controlled format. It supports cross-modal queries and can stream data straight into training, and its open-source core can be self-hosted or run as a managed service. The newer Deep Lake PG unifies a serverless Postgres with its vector and tensor engine.

Native vector storage and querying in S3 — serverless, billion-vector scale.

Purpose-built vector storage inside Amazon S3: store and query up to two billion vectors per index across thousands of indexes per vector bucket, with S3's durability and elasticity and no clusters to manage. Frequent queries return in around 100ms and infrequent ones in under a second, and it plugs directly into Amazon Bedrock Knowledge Bases for RAG.

Open-source memory for AI agents.

An open-source semantic memory layer for AI agents. Cognee ingests documents, relational data, and system context, then runs an Extract-Cognify-Load pipeline that uses an LLM to build a knowledge graph with embeddings and relationships. Agents query it for durable, cross-session context that captures how concepts connect. Self-host the Python SDK for free, or use the managed cloud tiers.

Open-source serving engine for vector, lexical, and structured search at scale.

A big-data serving engine that combines approximate nearest-neighbor vector search, lexical search, structured filtering, and ML model inference in a single query, evaluated over data distributed across many nodes. Battle-tested at Yahoo scale, it is offered as a free engine you self-host, or as the managed Vespa Cloud — including an Enclave mode that runs inside your own AWS or GCP account.

Distributed vector database built for billion-scale search.

Cloud-native, Apache-2.0 vector database for similarity search at scale, powering RAG, semantic and multimodal search, and recommendations. Its distributed architecture separates storage and compute and supports many index types (HNSW, IVF, FLAT, DiskANN, SCANN) with quantization and mmap. Created by Zilliz, which offers the managed Zilliz Cloud.

Embedded multimodal vector database on the Lance format.

An open-source retrieval engine for AI built on the Lance columnar format. It runs in-process alongside your app — no separate server — and stores, indexes, and searches vectors, metadata, and multimodal data (text, images, video) with vector, full-text, and SQL queries. A managed enterprise lakehouse tier scales the same engine to petabytes.

Vector similarity search inside Postgres. The pragmatic default.

Postgres extension that adds a vector type plus exact and approximate nearest-neighbour search. Pairs naturally with Supabase, Neon, and any managed Postgres. The lowest-friction RAG backend if you already run Postgres.

Fully-managed serverless vector database for RAG and semantic search.

Fully-managed vector DB built for production RAG and semantic search at scale. Serverless pricing, low-latency reads, and integrations across every major framework, with no infrastructure to provision or operate.

Postgres-based backend with auth, storage, and edge functions.

A common default backend when an app needs persistence + auth + realtime. Open-source, self-hostable, very low friction to local dev with the CLI.

Object-storage-backed vector DB. Serverless economics at scale.

Bills like S3 — cold rest, warm reads, no per-namespace minimums. Designed for very-large, mostly-cold vector workloads where you can't justify keeping every index in RAM. Operated by Notion in production.

Open-source, Rust-based vector DB. Fast, predictable, self-hostable.

Vector database written in Rust with a strong focus on filtering, payloads, and predictable latency at scale. Self-host on a single binary or use the managed cloud.

Open-source vector database with built-in vectorisers.

Cloud-native vector DB that can compute embeddings inline — pass raw text in, store vectors out. Strong hybrid (BM25 + vector) search; self-hostable or managed.

Embedded vector DB. Pip-install, prototype, scale later.

The low-friction starting point — Chroma runs embedded inside your Python process or as a hosted service. Great for prototypes and small-to-medium RAG apps; upgrade to a managed option when you outgrow it.