All-solid-state Li metal batteries (ASSLBs) are coming with sulfide solid-state electrolytes (S-SSEs) for superior
Li$^+$ conductivity, but irregular particles and interfaces lead to disorder Li+ flux in S-SSEs that hinder pure Li as an
anode. Specially, its mesoscopic structure cannot be adequately described by average size, making it difficult to analyze Li$^+$ flux effectively. Herein, a model is constructed on the molding of Li$_{5.5}$PS$_{4.5}$Cl$_{1.5}$ (LPSC) particles and defined size as the number (N) and consistency (σ ) to evaluate their effects on Li$^+$ transfer and concentration uniformity. Through machine learning of calculation data (Li+ concentration with N and σ ) and experimental results, excessive interfaces can hinder Li$^+$ transport and local aggregation of irregular interfaces leads to uneven ion transport. Therefore, a particle size gradient S-SSEs (induced by different size LPSC particles) is predicted to achieve fast and uniform Li$^+$ transport. Subsequently, this designed S-SSE is applied in ASSLBs, which can complete a 1000 h cycle with capacity retention exceeding 80%. This study elucidates that the long cycle ASSLBs can be achieved by adjusting the molding of LPSC particles. Specifically, it demonstrates that the Li$^+$ flux of the whole S-SSEs can be optimized through gradient size design.