Skip to content

All pages

Router

Each prompt gets routed to a model in under 50 milliseconds.

AIJul 11, 2026
Category
AI
Website
github.com
Language
English
Listed
Jul 11, 2026
Forks
127

weave-os/router ↗, Go, Last commit 15h ago

Router decides which model a prompt should go to in under 50 milliseconds. It sits in front of an agentic system and picks the right model per request, which moves the wiring between models out of application code and into one place. Adoption is a matter of pointing at a different endpoint, and the project puts the resulting savings at 40 to 70 percent. That 50 millisecond budget is the constraint it holds itself to, since a router that thinks too long becomes the latency it was meant to save.

Like this product?