signing its first contract with a frontier lab for tokens for intelligence to uh renting its own inference has dropped from 24 months uh about a year ago to 5 months today. And so we're really seeing people kind of run down the stack of frontier model, oops, this is too expensive. I need to route between different AI models. Oops, this is still expensive. I'd like to rent my own inference and open router plus MPP is perfectly positioned then for the
I need to route between different AI models.