Key Info
Requesty announced a beta feature called Auto Routing that lets teams describe the kinds of work they do, point each kind at a model, and have Requesty choose the model on the first message of a session.
Highlights
- The model choice is pinned for the entire session, so the model never switches mid-task and prompt caches keep working.
- If the classifier is unsure, it falls back to the model the team trusts most.
- Requesty reports its own Claude Code sessions on Auto Routing were 42% cheaper and 30% faster than running everything on Opus 5.5.
- Teams can set up a routing policy through the provided link.
