you have to be InferenceMAX™ing. you need to be efficiently allocating compute. you need to be using open-source tools to understand true cluster tco. you need to be emulating real-world applications. congrats to @dylan522p and @SemiAnalysis_ team