The conformable fractional GM(1,1) (CFGM(1,1)) model, which is based on the conformable fractional accumulation, is a recently proposed univariate grey prediction model. It is shows in this paper that the conformable fractional accumulated generating operator does not satisfy the new information priority principle which is extremely important in grey prediction theory. And the CFGM(1,1) model suffers from transformation errors from difference equation to differential equation. To address these two issues, a new information priority conformable accumulated generating operator is proposed. By combining the idea of the discrete GM(1,1) model, a new information priority conformable fractional discrete GM(1,1) model is constructed. The conditions for the novel operator to satisfy the new information priority principle are theoretically derived. Meanwhile, the quantum particle swarm optimization algorithm is adopted to determine the optimal accumulation parameters. Two practical examples demonstrate that the proposed model not only satisfies the new information priority principle but also effectively overcomes the transformation errors in the CFGM(1,1) model, resulting in better fitting and prediction accuracy.