The almost trivial answer is silica glass - it has close to zero expansion coefficient so it will not get stressed significantly on cooling.

What may be more important is the content of the crucible (assuming you don't just heating and cooling empty crucibles for fun). If you have a melt solidifying on cooling somewhere below 1260C, and that (as is likely) has a much higher thermal expansion coefficient than the crucible material, the fate of the crucible will depend largely on the quantity of solidifying melt and its adhesion to the crucible. Large amount, strong adhesion -----> cracked crucible.

Emanuel Cooper thank you very much for the suggestion. Indeed I would not use an empty crucible since the crucible will be filled with a homogenized mixture of limestone and clay and sinter at 1260 with no melt formation.