This paper proposes a novel learning-augmented deterministic and truthful mechanism for the strategic unrelated machine scheduling problem using fewer predictions to achieve the best possible consistency and robustness.