Something about @OpenGradient's approach feels different. The focus on making AI inference trustless rather than just fast is a subtle but important distinction. $OPG #OPG