Vector databases
Each provider is profiled from its own documentation, with the source recorded against every figure. Dimensions and similarity metrics decide whether an embedding model's output fits at all; hybrid search decides whether keyword relevance survives the move to vectors.
| Provider | Entry paid plan | Max dimensions | Hybrid search |
|---|---|---|---|
| Elastic Cloud | Not documented | Not documented | Supported — combines vector similarity scores with BM25F text-relevance scores in a single hybrid query. |
| Pinecone | Builder plan: $20/month flat, for solo developers and small teams. Standard plan is usage-based with a $50/month minimum. | Not documented | Supported — combine a keyword signal (BM25 or sparse) with a dense signal at query time, often with reranking. |
| Qdrant | Not documented | Not documented | Not documented |
| Redis Cloud | Essentials: from $0.007/hour, $5/month total. 250 MB-100 GB RAM and SSD, single DB. | Not documented | Not documented |
| Turbopuffer | Launch plan: $16/month minimum usage, including all database features. Scale plan: $256/month minimum. Enterprise: $4,096+/month. | Not documented | Supported — combines vector search and BM25 full-text search to produce semantically relevant results. |
| Weaviate | Flex tier starts at $45/month, pay-as-you-go, no commitment. | Not documented | Supported across all plans. |
Which to consider
Teams that need full-text search and observability on the same underlying engine, or that want the option to self-host later — Elastic's dual identity as a search platform and an observability platform is the actual differentiator against single-purpose competitors.
Teams that want a fully managed vector search product with hybrid search and metadata filtering built in, and don't want to operate the underlying index themselves.
Teams that want a choice of similarity metric rather than one fixed default, or that want the option to self-host the same open-source engine the managed cloud runs.
Teams that want Redis from the company that maintains it, with an on-demand hourly Essentials tier cheap enough to run a small production workload without committing to a fixed plan.
Teams that want vector search billed like storage rather than compute — object-storage-backed pricing tends to undercut memory-resident vector databases at rest, especially for large, cold datasets.
Teams that want hybrid search included on every tier without an add-on, or that want an always-free cluster (not a time-limited trial) to build against before committing to a paid plan.