The Doppler Orbitography and Radiopositioning Integrated by Satellite (DORIS) technique is one of the four space-geodetic techniques contributing to the International Terrestrial Reference System (ITRS) and its recent realizations, including the extension ITRF2020-u2024. Within the International DORIS Service (IDS), the IGN-IPGP/JPL Analysis Center routinely estimates tropospheric parameters when processing radiometric measurements. In this study, we analyze the internal consistency of DORIS-derived tropospheric products from a reprocessing of multi-satellite data over 2021.0–2025.0 with the IGN22 configuration. Tropospheric delays at DORIS beacons are modeled by estimating ZWD and horizontal gradients, while VMF1-based mapping functions and VMF1 zenith hydrostatic delays are used as a priori information. We consider two processing strategies, FIXED and FREE, which differ by the strength of the station constraints while sharing the same tropospheric parameterization. Their stability is assessed using internal consistency metrics based on day-to-day and multi-satellite overlaps at common beacons. The comparison between both solutions shows that ZWD estimates are highly robust to the station-constraint strategy, with a median absolute FIXED–FREE difference of 2.9 mm and negligible global bias. Day-to-day overlap statistics indicate median absolute inter-day differences of 3.2 mm for FIXED and 4.2 mm for FREE. Inter-satellite diagnostics further show that mono-satellite ZWD estimates are generally consistent at the millimeter-to-centimeter level, with the largest improvement obtained when the number of simultaneously visible satellites increases from two to three. These results establish a first internal benchmark for the stability of DORIS tropospheric products in the IGN22 configuration and provide a basis for future external validation against GNSS- and model-based tropospheric references.