Predicting solution gradients (velocity gradients, pressure gradients, shear stresses, heat flux, etc.) are critical in many applications and across the entire flight regime, Reynolds numbers, and Mach numbers. These parameters play a vital role in both the accuracy and the validity of the performed analysis, design and uncertainty quantifications, and not only provide a significant challenge in deigning numerical algorithms, they also impose serious limitations on the quality of the grids, the type of the elements used in certain regions of flow field, and the grid adaptations, particularly for unstructured grids. We briefly look at a few real examples where some of the above mentioned limitations are observed. We overview some of the recent advances made and/or attempted in addressing these issues. In particular, we look at an overall reformulation strategy of the target governing equation such that the gradients of the primal variables become independent variables. We illustrate that such approach appears to significantly improve the predicted solution gradients without limitations on the grid quality. We finish the talk, we some remarks and suggestions for future work.