Anisotropic fast-marching algorithms are computationally efficient tools for generating realistic maps of karst conduit networks, constrained by both the spatial extent and the orientation of karstifiable geologic units. Existing models to generate conduit network maps are limited either by high computational requirements (for chemistry-based models) or by their inability to incorporate the effects of elevation and orientation gradients (for isotropic fast-marching models). The new anisotropic fast-marching approach described here provides a significant improvement, though it imitates rather than reproduces actual speleogenetic processes. It can rapidly generate a stochastic ensemble of plausible networks from basic geologic information, which can also be used as input to karst-appropriate flow models. This paper introduces an open-source, easy-to-use implementation through the Python package pyKasso, then describes its application to a well-mapped geologically complex long-term study site: the Gottesacker alpine karst system (Germany/Austria). Groundwater flow in this system is exceptionally well understood from speleological investigations and tracer tests. Conduit formation primarily occurs at the base of the karst aquifer, following plunging synclines. Although previous attempts to reproduce the conduit network at this site yielded implausible network maps, pyKasso quickly generated networks faithful to the known conduit system. However, the model was only able to generate these realistic networks when the inlet-outlet connections of the system were correctly assigned, highlighting the importance of pairing modeling efforts with field tracer tests. Therefore, a model ensemble method is also presented, to optimize field efforts by identifying the most informative tracer tests to perform.