LoginSign Up
← Back to all sparks
W

Weaviate

DEVOPS
Velocity5.0

Open-source vector database for AI applications and semantic search

weaviate.io ↗

Weaviate graduates search APIs to GA while tutorial volume signals an engine in maintenance mode

vector-databaseragmulti-vector-retrievalsearchdeveloper-tools
◆Current state
Weaviate 1.39 promoted the Boost API and MMR diversity selection to general availability, previewed 4-bit Rotational Quantization — a significant memory-reduction path for large deployments — and shipped an experimental Search REST API. The Query Agent gained three effort tiers (medium, high, ultrahigh) for test-time compute scaling. Between releases, the team publishes developer tutorials at high frequency, including a multi-part Foundry series on building semantic search into creative workflows.
◆Where it's heading
The core vector-search engine is stabilizing: recent releases promote experimental features to GA rather than introducing new primitives. Memory efficiency (4-bit RQ) and the Query Agent effort-tier model are the active development fronts. An earlier free-to-start Cloud pricing shift signals a volume acquisition play on top of this foundation.
◆Prediction
4-bit Rotational Quantization moving from preview to GA is the most likely next release milestone. The Foundry tutorial series suggests a template or starter kit targeting creative and media teams is being assembled.

◆Recent moves

  1. 7d ago

    HFresh: Memory-Efficient Vector Search

    View source ↗
  2. 8d ago

    Building Foundry Part 3: From archive to creative search

    Part 3 of the Foundry tutorial series — demonstrating how to build a searchable creative archive using a manifest, Weaviate, and hybrid search. Developer education content, not a product release.

    View source ↗
  3. 15d ago

    How to extract meaning from charts and tables in PDFs

    A technical tutorial on using late-interaction multi-vector retrieval to search PDF charts and tables without OCR or text extraction. Demonstrates a real Weaviate capability but is developer education, not a product update.

    View source ↗
  4. 20d ago

    Weaviate 1.39 Release

    Weaviate 1.39 graduates the Boost API and MMR diversity selection to GA, adding stability guarantees for two widely used search features. The 4-bit RQ preview and experimental Search REST API are the new surfaces — both early availability but signaling where memory efficiency and API design are heading.

    View source ↗
  5. 1mo ago

    Building Foundry Part 2: Where creative workflows break

    Part 2 of the Foundry series, covering why folders, tags, and keyword search break down in real creative workflows and what retrieval needs to do instead. Developer education, not a product release.

    View source ↗
  6. 1mo ago

    Scaling Test-Time Compute in Search Mode

    The Query Agent's Search Mode gains three effort tiers — medium, high, and ultrahigh — letting callers trade latency for retrieval quality. A meaningful addition for precision-sensitive use cases, fitting the trajectory toward a more configurable agentic search layer.

    View source ↗