I will present the evaluation of a large suite of dark energy and modified gravity models. By the use of a p-value test instead of some traditional model selection methods, we find that some models can be ruled out by the current observational data. We also explore the possibilities of the galaxy clustering to detect some CMB-like anomalies. In particular, we perform a Fisher matrix analysis for the ongoing galaxy survey. The result shows that the DESI galaxy sample can detect the hemispherical power asymmetry first reported in the CMB observation with a significance higher than 3\sigma. We also develop an emulator approach for the modeling of the galaxy clustering at small scale. The resulting product can provide the real space clustering and redshift space multipoles with an accuracy better than 1%, which also allows the exploration of the constraints on cosmology.