Weaviate Query Agent adds effort tiers to balance accuracy, latency and cost

CShorten30 · x · 2026-09-29

Weaviate introduced an effort parameter for Query Agent Search Mode with medium, high and ultrahigh tiers, letting users control how much work the agent puts into each request.

Higher effort scales computation in two places: query writing (longer decomposition into structured searches) and reranking (more thorough evaluation of retrieved documents). Weaviate recommends testing tiers against your own accuracy, latency and cost requirements, since the right tier depends heavily on the workload.

Original post →

More from coding & agent

coding & agent channel →