A metal fuel performance code was coupled to a subchannel analysis code to predict, in a computationally efficient way, critical phenomena that could lead to pin failure for steady-state and transient scenarios in sodium-cooled fast reactors. The fuel performance and subchannel codes coupled are FEAST-METAL and an updated version of COBRA-IV-I, respectively. In coupling the codes, the importance of azimuthal temperature and stress effects in the fuel pin were analyzed; it was concluded that azimuthal temperature averaging around the fuel pin is an acceptable approximation. The codes were coupled using a wrapper, the COBRA And FEAST Executer (CAFE), written in the Python programming language. Data from EBR-II was used to confirm and verify CAFE. Finally, CAFE was used to predict the maximum allowable burnup of three different fuel assembly designs (driver fuel, radial blanket, and tight-pitch breed-and-burn fuel) as a function of operating temperature, linear power, fuel composition, cladding thickness, and smear density.