Herein, we publish the simulated global annual mean sea surface temperature (THO), surface air temperature (SAT) over a time period of 100 years retrieved from equilibrium climate simulations for the Last Glacial Maximum (~21 ka BP). We investigate the range of temperature variability that occurs in response to uncertainties in the boundary conditions of Laurentide ice sheet (LIS). We performed LGM simulations, applying six different LIS reconstructions (ICE-6g, GLAC-1a, ANU, Gowan, Licciardi and PMIP3) in a fully coupled atmosphere-ocean-sea-ice model. The model data has been used in the publication by Hossain et al., 2021. The climate data has been produced with Consortium for Small-scale Modeling (COSMOS; ECHAM5/JSBACH/MPIOM/OASIS3), utilized at a resolution of T31 in the atmosphere with 19 vertical layers and a resolution of GR30 (~3.0°x1.8°) in the ocean with 40 vertical layers. The model setup refers to boundary conditions (terrestrial topography, ocean bathymetry), greenhouse gas concentrations (CO2 = 185 ppm; CH4 = 350 ppb; N2O = 200 ppb) and orbital forcing representative for the LGM and are imposed in accordance with the PMIP3 protocol. We also run COSMOS using PI boundary conditions (ice-sheet topography, orbital forcing, greenhouse gas concentrations and ocean bathymetry). Details on setup and identifiers of LGM model simulations can be found in Table S1 of Hossain et al., 2021.