Weaviate
Open-source vector database with hybrid search and built-in embedding modules.
why this verdict
Downgrade — pay for the workflow, not the AI
pgvector and Qdrant cover the headline feature free, so the $25/mo is now paying for the workflow around it.
- A named launch, not a vibe named, unlinked
- Pgvector making a dedicated vector database unnecessary for most retrieval workloads. OpenAI, April 16, 2025 — no announcement link recorded yet.
- How much of the job it covers headline only editorial call
- The headline feature, but not the workflow, data or integrations around it.
- Is there a free way to do it? yes
- 2 of 3 listed replacements have a usable free tier: pgvector and Qdrant.
- What the call is worth $300/yr
- $25/mo at the entry paid tier — $300 a year per seat.
- Threatened by
- OpenAI
- Since
- April 16, 2025
- List price
- $25/mo
- Per year
- $300
The backstory
Weaviate was among the first to make vector search approachable, bundling embedding generation, hybrid keyword-plus-vector retrieval and a graph-ish schema into one server. Then pgvector got fast enough that most teams stopped adding a database to their stack for retrieval, and long-context models reduced how much retrieval some applications need at all. Weaviate is still the better tool past a certain corpus size, but the number of teams who ever reach that size shrinks every time Postgres improves.
Escape hatches
switching could free up $300/yrVector search inside the Postgres you already run
github.com open_in_newRust vector database with strong filtering performance
qdrant.tech open_in_newObject-storage-backed vector search at much lower cost
turbopuffer.com open_in_new2 of 3 replacements have a usable free tier.
