Weaviate is an open-source vector database, available self-hosted or as Weaviate Cloud, with a genuinely free tier (100,000 objects, 1GB memory) and pay-as-you-go Flex starting at $45/month. Pick it when you need a standalone vector database server (not an embedded library) for similarity search over your own documents or records. Plan how embeddings get produced and kept in sync before committing to a schema.
Other good fits
Teams that want a standalone database server (with replication, RBAC and uptime SLAs on paid tiers) rather than an embedded library.
+ Pros
Read pros
Weaviate is an open-source vector database for storing and searching embeddings (numeric representations of text, images or other content) to power similarity search and retrieval-augmented generation. It's available self-hosted (open source) or as Weaviate Cloud, a managed service built on the same open-source project, with a genuinely free tier for getting started.
Pick it when…
You want to start a vector search project on a free cloud tier before committing to infrastructure spend.
You need a standalone database server, with features like replication and role-based access control on paid tiers, not an embedded library inside your own process.
You want the option to move between self-hosting the open-source project and a managed cloud service without changing the underlying technology.
− Cons
Read cons
A vector database doesn't generate embeddings for you from nothing; you still need to choose an embedding model and keep embeddings in sync as your source data changes, which is real ongoing work, not a one-time setup step. The free tier's 100,000-object and 1GB-memory limits are real constraints, not just a trial; plan your migration to Flex before you hit them, not after.
Pick something else when…
You want an embedded library with no separate server to run.LanceDB runs inside your own process instead.
Your retrieval problem is really about relationships, not similarity.Neo4j is built for graph traversal rather than vector search.
You want to prototype the retrieval pipeline visually first. A visual flow tool can help you decide on your embedding and retrieval approach before committing to a specific database.
Avoid if
Your dataset will exceed the free tier's 100,000 objects and 1GB memory quickly; budget for Flex or Premium from the start.
Workflow
Define a collection (schema) for your data → generate embeddings for your records using an embedding model of your choice → store records and their embeddings in Weaviate → query with a vector or hybrid (vector + keyword) search → inspect the returned records against your actual question → adjust the schema or embedding approach if results aren't relevant.
Shared cluster, full core DB toolkit, replication, RBAC, 99.5% uptime
Premium
Custom, prepaid contract
Shared or dedicated deployment, up to 99.95% uptime, multi-cloud (AWS/GCP/Azure), enterprise support
Self-hosting the open-source project is free beyond your own infrastructure cost, since Weaviate Cloud is described as the same open-source project wrapped in a managed service; specific self-hosted infrastructure sizing wasn't detailed on the pricing page checked. Running effort includes deciding how embeddings get generated and kept in sync as your underlying data changes, which is a real ongoing responsibility regardless of hosting choice.
Restrictions
Self-hosted open-source infrastructure sizing isn't detailed on the pricing page checked; size your own deployment based on your dataset, not the managed-cloud tier limits. Free tier resource caps (100k objects, 1GB memory) are enforced, not advisory.
First project
Documentation-based suggestion — not run here.
Read first project
Build a small semantic search over your own notes.
Sign up for Weaviate Cloud's free tier (100,000 objects, 1GB memory is plenty for a first project).
Create a collection and add 15-20 short text records (your own notes or documents), letting Weaviate's built-in embedding integration or your own embedding model handle vectorization.
Run a query using natural language, not exact keyword matches, and check whether the top results are actually relevant.
Try a hybrid search (combining vector and keyword) on the same query and compare results.
Done when a natural-language query correctly surfaces the most relevant record from your set, verified by reading the actual returned text.
Links & evidence
Documentation-based · not hands-on · Sep 23, 2026
Rests on Weaviate's pricing page, homepage and documentation, checked 23 September 2026. No collection was created and no query was run for this review.
Weaviate pricing · Sep 23, 2026 — Free/Flex/Premium tiers, object and memory limits, uptime SLAs, self-hosted vs. cloud framing.
Store objects with embeddings for semantic and hybrid search.
Build retrieval-augmented generation queries through GraphQL or REST.
Run locally, embed it in Python or JavaScript, or use managed cloud hosting.
free-tier · cloud · self-host
vector database
semantic search
hybrid search
rag
embeddings
search-cite
open-source
View sourced imageInterface frame captured from the Weaviate Cloud Explorer walkthrough embedded in its official documentation.
Weaviate official product documentation
Owner-requested editorial use of public official interface screenshots, 15 September 2026. Copyright retained by the original publisher; attribution preserved. No open licence or hands-on course execution is asserted. Source asset: https://app.guideflow.com/embed/5pvnnvxtwk. PNG conversion preserves the complete frame.