Looking at the Capital Cost variables with a specific focus on solar PV (Capital Cost|Electricity|Solar|PV), my impression is that there are several issues related to regional aggregation across different models that report these variables.
AIM/CGE 2.1: The value reported under World region appears to be the sum of the values reported at the 5-regional level.
REMIND 1.7/REMIND-MAgPIE 1.7-3.0/MERGE-ETL 6.0/GENeSYS-MOD 1.0: The values reported at the 5-regional level appear to be multiples of the value reported under World. It seems that the number of native regions model regions per "5-region" determines the factor.
IEA Energy Technology Perspective Model 2017, IEA World Energy Model 2017 and IMAGE 3.0.1 don't seem to report World aggregates, so it is harder to diagnose whether the same holds for these models, but it would make sense to double check this.
Looking at the Capital Cost variables with a specific focus on solar PV (Capital Cost|Electricity|Solar|PV), my impression is that there are several issues related to regional aggregation across different models that report these variables.