@ZyMazza much as i like local models i also feel like letting the market decide is best wrt gpu & ram allocation, i expect the utilization is higher in a datacenter than at home which means cheaper tokens overall