ManyVector

What ManyVector is

ManyVector is a vector database built on object storage, with vector, full-text, and hybrid search for applications such as RAG pipelines, semantic search, recommendations, and AI agent memory. Its HNSW approximate nearest-neighbour search advertises sub-10ms p50 latency and 90%+ recall@10. BM25 supports full-text search, while hybrid search combines vector and text scores using RRF and metadata filters. Copy-on-write namespace branches provide snapshots, evaluation branches, and rollback without copying vector data.

ManyVector supports AWS S3, Google Cloud Storage, Cloudflare R2, Azure Blob Storage, MinIO, Tigris, Wasabi, and other S3-compatible stores. It says it works with OpenAI, Cohere, and any embedding model or provider. Python and TypeScript SDKs are included in the Free Tier. ManyVector says vectors and indexes remain in the customer's object store and customer data is not stored on its disks. For managed SaaS, it states that data at rest is encrypted with AES-256 and client traffic uses TLS 1.2 or higher. ManyVector is a product of ManyLayers.

Who it suits

ManyVector suits developers building search or retrieval features who want vector, text, or hybrid retrieval and can bring an object store bucket. Its listed use cases include RAG pipelines, semantic search, recommendations, and AI agent memory. Organizations needing private deployment, identity controls, or an uptime SLA can consider Enterprise.

Limits

The Free Tier allows 100M search queries per month and requires your own object store bucket. Growth includes 100M free queries per month and bills for queries above that amount.

Getting started

1. Visit the ManyVector website.
2. Choose the Free Tier, Growth, or Enterprise plan.
3. For the Free Tier, bring an object store bucket.
4. Use the Python or TypeScript SDK included in the Free Tier.

Verdict

Choose ManyVector if you need vector and text retrieval with data stored in your object store, plus branching for snapshots and rollback. The Free Tier includes up to 100M search queries per month and Python and TypeScript SDKs. Look elsewhere if you need a published Enterprise price; the plan is contact for volume pricing.