latest dive · no. 6· 43 min
token maxing.
token maxing. Sam and Steve on running out of usage, cheaper models behind good review gates, going from tasker to guide, local models catching up to the labs, and why looping agents is like leaving the house with the lights on.