Diketone and aryl hydrazine condensation reactions often lead to a mixture of regioisomeric pyrazoles. Preliminary calculations show 2-4 kcal/mol differences in the pyrazole energies.
if your TS or reaction intermedia do not show radical character a nice DFT calculation using a "classical" B3LYP/6-31G(d,p) level of theory is enough. I presume that to better analyze the subtilities of your energy difference you want to perform also a vibrational calculation (anyway you have to do it to check if your TS is a real TS) and possibly include a solvation model (also a simple and elegant COSMO) to see the difference in the free energies and not only in entalpies or SCF.
However nowadays there are very efficient hybrid functionals like the B2PLYP though requesting a 2-5 fold computational power the results are really reliable.
Obviously thej basis set depends from your cluster. A triple zeta like a 6-311G(d,p) (or higher) are always welcome