I would check what the updates are from one version to the other.Very unlikely, but maybe they tweaked default values like convergence thresholds, smearing types, k-point grids, diagonalization algorithms, etc. You can also check the output files from both runs to see if you can spot any differences in parameters.

Rgds

Gabriel Vinicius

It was not the version; the initial output file from one version was not optimized, so that was making a difference. Although slightly there is variation in cell parameters and total energy and little in band gap (negligible).

Thank you