Blog Post

Microsoft Foundry Blog
5 MIN READ

Let the query decide: introducing auto retrieval reasoning effort

amaias's avatar
amaias
Icon for Microsoft rankMicrosoft
Sep 30, 2026
Introducing auto : a new retrieval reasoning effort mode that recovers 97.6% of the evidence recall of the highest effort level at 46.9% fewer query planning tokens. Enterprise retrieval workloads...
Average evidence recall and query-planning tokens by retrieval reasoning effort.
Updated Sep 22, 2026
Version 1.0
PradyH's avatar
PradyH
Icon for Microsoft rankMicrosoft
Oct 27, 2026

Although this feels conceptually similar to Model Router, but applied to retrieval, I think of it as dynamically selecting the retrieval effort rather than the model. Hope that makes sense.