Aluminium (Al 99.99 % purity) and aluminium alloy Al-5.7Ni and Al-5Mg (wt.%) open-cell foams made by replication processing are creep-tested in the temperature range T/Tm=0.5-0.8, varying the relative density between 0.10 and 0.25. Steady state creep rates are measured and interpreted on the basis of analytical models developed using the variational estimate of Ponte Castañeda and Suquet, as well as the beam engineering model developed by Andrew-Gibson-Ashby. The steady state creep rate, in foams produced of pure aluminium show a strong sensitivity to the volume fraction: for instance, in the case of the 400 µm pore size pure Al it varies as. The present experiments are therefore conducted on foams produced of two metals having a substructural scale much smaller than the pore size (the Al3Ni rod spacing and the average dislocation spacing, for Al-5.7Ni and Al-5Mg respectively), with a goal to probe whether this behaviour should be attributed to the material constituting the foam (microscopic scale) or to the foam architecture (mesoscopic scale). The creep behaviour is examined in light of the pure aluminium data, and interpreted in terms of the possible influence of the damage, the presence of internal stress or the substructure-invariant model of Sherby et al.