Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

ManyVector

manyvector.com

  • Free plan
  • API
  • Self-hosted
6.8 Club score Visit manyvector.com

What ManyVector is

ManyVector is a vector database built on object storage, offering vector, full-text, and hybrid search. It uses HNSW approximate nearest-neighbour search and advertises sub-10ms p50 latency with 90% or higher recall@10. Full-text search uses BM25, while hybrid search combines vector and text scores through reciprocal rank fusion with metadata filters. Copy-on-write namespace branching supports snapshots, evaluation branches, and rollback without copying vector data. Supported storage includes AWS S3, Google Cloud Storage, Cloudflare R2, Azure Blob Storage, MinIO, Tigris, Wasabi, and S3-compatible stores. ManyVector says it works with OpenAI, Cohere, and other embedding providers. Listed uses include RAG pipelines, semantic search, recommendations, and AI agent memory. Vectors and indexes are stored in the customer’s object store, according to the maker, rather than on its disks. Managed SaaS data at rest uses AES-256 encryption and client traffic uses TLS 1.2 or higher. The Free Tier includes Python and TypeScript SDKs, 100 million monthly search queries, and community support. Growth includes 100 million free monthly queries with per-query billing above that allowance. Enterprise adds self-hosting, SSO/SAML, SCIM, and a 99.9% uptime SLA.

Who it suits

ManyVector suits teams building retrieval, semantic search, recommendation, or agent-memory systems around object storage. It can also fit organizations needing self-hosted or air-gapped Enterprise deployment.

Good to know, and what to watch

  • Vector, BM25, and hybrid search
  • Copy-on-write namespace branching
  • Many object-storage integrations
  • Python and TypeScript SDKs
  • Customer-controlled object-store data
  • Free Tier requires bringing an object-store bucket
  • Growth charges per query above its allowance
  • Some deployment and identity features require Enterprise

Enterprise

—

Free Tier

Free

Growth

Free

Free plan · paid from $0.00

The verdict

ManyVector combines search modes, branching, and customer-managed object storage for AI data workloads. Its free allowance supports experimentation, while self-hosting and governance options are reserved for Enterprise.

What the maker publishes

Compliance
The maker states that its managed SaaS platform is SOC 2 Type II certified and offers BAAs on Enterprise plans. source
Data handling
ManyVector says vectors and indexes are stored in the customer's object store and it does not store customer data on its disks. source
Embedding compatibility
ManyVector says it works with OpenAI, Cohere, and any embedding model or provider. source
Full-text and hybrid search
It offers BM25 full-text search and hybrid search combining vector and text scores with RRF and metadata filters. source
Maker
ManyVector is a product of ManyLayers. source
Namespace branching
Copy-on-write namespace branching supports snapshots, evaluation branches, and rollback without copying vector data. source
Product
ManyVector is a vector database on object storage with vector and full-text search. source
SDKs
The Free Tier includes Python and TypeScript SDKs. source
Security
The security page states managed SaaS data at rest is encrypted with AES-256 and client traffic uses TLS 1.2 or higher. source
Self-hosting
Enterprise customers can deploy in their own VPC or on-premises, including air-gapped operation after initial setup. source
Storage integrations
Supported object stores include AWS S3, Google Cloud Storage, Cloudflare R2, Azure Blob Storage, MinIO, Tigris, Wasabi, and S3-compatible stores. source
Support
Free Tier includes community support, Growth includes priority email support, and Enterprise includes dedicated support. source
Use cases
The maker lists RAG pipelines, semantic search, recommendations, and AI agent memory as use cases. source
Vector search
It uses HNSW approximate nearest-neighbour search and advertises sub-10ms p50 latency with 90%+ recall@10. source

Checked against the maker's pages on 5 Oct 2026.