Meta dropped three new models over the weekend: Scout, Maverick, and the still-training Behemoth, billed as the next evolution of “open-ish” AI. But instead of excitement, the response was mostly shrugs. Critics called the release underwhelming, saying it lacked the edge expected in today’s breakneck AI race. Meta’s clear attempt to claw back some attention quickly turned messy. Accusations began circulating on X and Reddit around benchmark tampering, a mystery ex-employee, and large gaps between the models’ public and private performance.








