The Istanbul GPS Triangulation Network (IGTN) and the Istanbul Levelling Network (ILN), established in 2006, provide data for the determination of a local GNSS/levelling geoid model. These networks' measurements were done separately on both the Asian and European sides of the Bosphorus Strait in the vicinity of Istanbul. To connect these regions for those networks, a Valley Cross Levelling (VCL) data set, acquired in 1986 and 2004, was used. The use of this VCL data set was challenging in calculating the Istanbul geoid model, primarily because of its errors. In this study, this challenge was overcome through newly collected VCL data in 2010, allowing for the readjustment of the ILN and the newly collected VCL data set. The Istanbul geoid model was computed using soft computing techniques including the adaptive-network-based fuzzy inference system (ANFIS) and the artificial neural networks (ANNs). The resulting Istanbul GNSS/levelling geoid model is shown to be more reliable when compared with the model computed using conventional interpolation techniques. (C) 2020 Institute of Seismology, China Earthquake Administration, etc. Production and hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd.