Weaviate

What it is for

Weaviate is an open-source vector database: you store objects plus vectors, then hybrid-search or ask the Query Agent. Use it when the job is vectors you can self-host or rent from $45 Flex, not when you need a $20 chat seat.

What makes it unique: the cover is the official weaviate/weaviate GitHub file list — the OSS server — not a pricing card.

Key features

• Vector store — objects and embeddings in one collection, with HNSW / HFresh indexes and compression (pricing; GitHub). One database. Index type is SKU of this.
• Hybrid search — vector plus keyword in the same query (pricing comparison “Hybrid search”). A retrieval mode, not a second database.
• Query Agent — natural language → Weaviate operations (pricing Database AI Services). An agent door. Free 1,000 req/month; paid $30/org for 4,000 included.
• Weaviate Embeddings — hosted embed models inside Cloud (Arctic / ModernVBERT token rates). A model meter, not the DB.
• Self-host OSS — run the same server on your metal (weaviate/weaviate). A deploy. Cloud Free / Flex / Premium are SKU of the hosted path.

Clients and Docker Compose are install. Console “Query” is ops. Vector dimensions and GiB storage are the meter. Multi-tenancy is a mode of the store.

Who it's for

Best fit: teams who will stay on Cloud Free $0 (100k objects, 1 GB RAM, 10 GB disk) or self-host, then buy Flex at $45/month when they need HA, 1,000 collections, and pay-as-you-go dimensions.

Pricing

Verified 2026-10-22 (Asia/Shanghai) from weaviate.io/pricing. Cloud cards below. OSS is $0 to run yourself.

• Cloud Free — $0. Forever. 1 cluster / user. 100,000 objects. 1 GB memory. 10 GB disk. 1 collection, up to 3 tenants. Embeddings 2,000 req/day. Query Agent 1,000 req/month. Best-effort SLA.
• Flex — $45/month minimum, pay-as-you-go. Shared cluster, replication, RBAC, 99.5% uptime. Vector dimensions from $0.00465/1M. Storage from $0.12/GiB. Backup from $0.03036/GiB. 7-day backups. Query Agent free tier + usage.
• Premium Shared / Dedicated — prepaid, sales. Up to 99.95% (dedicated). SSO/SAML. HIPAA + PrivateLink on dedicated. Dimensions from $0.003875 (shared) / $0.002718 (dedicated). Minimum from $400/month on Premium (FAQ).
• Query Agent plan — $30/org/month, 4,000 requests included, then usage.
• Embeddings (list) — Arctic-embed-m-v1.5 $0.025/1M tokens; v2.0 $0.040; ModernVBERT $0.065.

Rates vary by index, compression, region. New price book from 1 September 2026 (RQ-4 + backup correction). Confirm the console Running Costs; taxes follow your location.

Limitations / costs

• Free is 100k objects and one collection. Flex is the first unlimited card.
• $45 is a floor. Dimensions, storage, and backups stack.
• A vector DB, not a notebook. For serverless DuckDB see MotherDuck. For a thin vector store see Turbopuffer.

Alternatives

• Qdrant — another OSS vector DB when you want Rust-first self-host.
• Chroma — a local embed store when you do not need Cloud HA.
• Turbopuffer — object-storage-first vectors when you do not want a Weaviate cluster.

Official tutorial

Official Weaviate vector database channel video (oEmbed author: Weaviate).

Written path: weaviate/weaviate plus Pricing.

Website

https://weaviate.io

Repository

https://github.com/weaviate/weaviate

Source

Catalog row curated for AI Tools Nav. Cover is the official weaviate/weaviate GitHub file list. Dollars checked on weaviate.io/pricing (Free $0; Flex $45/mo; Query Agent $30/org). Tutorial oEmbed author: Weaviate.

Last verified

2026-10-22 (Asia/Shanghai) editorial check.