-fitstat- and -estat ic- giving different AIC/BIC values

I am estimating a number of regressions, including OLS, tobit and GLM. Stata gives the AIC and BIC values when you do the GLM model, but for OLS and tobit there is an additional command. When I have tried to post estimate the AIC and BIC values for the OLS and tobit model I've noticed that the two commands -fitstat- and -estat ic- give different values for the AIC/BIC. which would be most appropriate to report.

I am not really using the AIC/BIC value for any reason, other than for presentation. Which command would give me the AIC/BIC value that is inline with what the GLM output gives?

Thanks,

Ash