Abstract: Objectives: Due to the generally lower quality of observation data from low-cost receivers, primarily characterized by a high occurrence rate of gross errors and frequent cycle slips in the measurements, fixing ambiguities for precise point positioning (PPP) is difficult. Furthermore, the positioning process often experiences repeated reconvergence. To address this issue, this paper proposes an improved IGGIII (Institute of Geodesy & Geophysics III) robust model. Building upon the classical IGGIII robust model, this improved model uses Kalman filter post-fit residuals to further analyze observation quality and incorporates cycle slip information to identify outliers. Methods: The improved IGGⅢ robust model was used to perform PPP partial ambiguity resolution experiments using static mode data collected by the SinoGNSS receiver in different scenarios. Results: The results indicate that the improved model significantly enhanced positioning accuracy in the horizontal direction in open environments. In occluded environments, it effectively improved positioning performance, reduced the time to first fix of ambiguity, and increased the ambiguity fix rate for both GPS (global positioning system) and BDS (BeiDou navigation satellite system), for both single-system and dual-system (GPS+BDS) PPP. Specifically, the improved model led to accuracy improvements of 16.81% (east) and 38.33% (north) for GPS, and 5.61% (east) and 1.39% (north) for BDS in single-system PPP. For the GPS+BDS dual-system PPP, the improvements reached 11.46% (east) and 35.75% (north). In occluded environments, positioning accuracy along the east, north, and up directions improved by 11.23%, 2.65%, and 23.14% for GPS, and by 14.96%, 12.35%, and 5.76% for BDS. For the dual-system solution, the respective improvements were 12.52% (east), 21.15% (north), and 8.21% (up). Additionally, the time to first fix of ambiguity was reduced by 26.05%, and the ambiguity fix rate increased by 13.61%. Conclusions: The improved model demonstrates enhanced performance in positioning accuracy and ambiguity resolution compared to the classical model. It proves effective in both open and occluded environments, especially in improving the robustness of single-system and dual-system PPP solutions. While convergence performance remains largely unchanged in open environments, the model shows notable advantages under challenging signal conditions.