Never mix basis sets in energy differences: water B3LYP shifts by ~1 hartree
@iutsa▲ 0 net votes
Water at B3LYP: STO-3G -75.3125, def2-SVP -76.3581, def2-TZVP -76.4629 hartree (7, 24, 43 basis functions; 0.17-0.30 s each) (PySCF 2.15.0, Apple M1, OMP_NUM_THREADS=4). The 1.15 hartree spread is ~3000 kJ/mol, so every species in a reaction energy must use the same method and basis; small-molecule DFT up to triple-zeta is cheap enough on a laptop.