This study aimed to identify the best model and method for the genomic selection of early growth traits in Inner Mongolian cashmere goats (IMCGs). Using data from 50,728 SNPs, the phenotypes (birth weight, BW; weaning weight, WW; daily weight gain, DWG; and yearling weight, YW) of 2256 individuals, and pedigree information from 14,165 individuals, fixed effects were analyzed using a generalized linear model. Four single-trait animal models with varying combinations of individual and maternal effects were evaluated using the ABLUP, GBLUP, and ssGBLUP methods. The best model was selected based on a likelihood ratio test. Five-fold cross-validation was used to assess the accuracy and reliability of the genomic estimated breeding values (GEBVs). Birth year and herd significantly affected BW ( < 0.05) and WW, DWG, and YW ( < 0.01), while sex, birth type, and dam age had highly significant effects on all traits ( < 0.01). Model 4, incorporating direct and maternal additive genetic effects, maternal environmental effects, and their covariance, was optimal. Additionally, ssGBLUP achieved the highest GEBV accuracy (0.61-0.70), outperforming the GBLUP and ABLUP methods. Thus, ssGBLUP is recommended for enhancing the genetic progress in IMCGs. Under the best method, the heritability estimates for BW, WW, DGW, and YW were 0.11, 0.25, 0.15, and 0.23, respectively.