One drawback of many nickel-base superalloys is their poor weldability. In particular, cracking can occur in the mushy zone during solidification, in the so-called Brittle Temperature Range (BTR). Another type of cracking is due to the formation of intergranular liquid films, by the liquation of low melting point phases like carbides and/or intermetallics. In this work, computational thermodynamics (Thermo-Calc) is used to predict the occurrence of these types of cracking, and to design Ni base alloys by minimizing both the BTR and the risk of γ’ liquation cracking, while keeping a good phase stability and good mechanical properties in the expected service temperature range. Optimized alloy compositions are determined by an automated maximization of stability and weldability, using genetic algorithms. Also, the creep rupture resistance is estimated through the regression of existing data, using neural networks. Predictions are then used to propose new weldable and creep-resistant superalloys.