DLR/GOSC started contributing precise clock offset corrections to the IGS real-time pilot project in late 2008 as one of the first real-time analysis centers with the Real-Time Clock Estimation engine RETICLE. The first version of RETICLE was a single-threaded GPS-only application. Later enhancements included the GLONASS constellation as well as the GIOVE test satellites. The deployment of new constellations and the modernization of the legacy systems lead to the need for processing many more satellites and signals, pushing the single-threaded processing to its limits. In mid-2015, the development work for the second version of RETICLE began, which has from the start been designed for processing a large network of reference stations and multiple GNSS constellations. The system relies on multi-threaded processing using a federated Kalman- filter design. For each station of the real-time network, the pseudorange, carrier-phase and C/N 0 observations are processed in a local station filter. Each of these filters parameterizes satellite clock offsets, one receiver clock offsets per constellation, the tropospheric wet zenith delay, differential code biases (DCBs), slant ionospheric delays and carrier-phase ambiguities. A global, merging filter then merges the satellite clock offset estimates from the local filters into a global solution. The result of the global clock combination filter is then fed back into the local filters. In a similar way, the local filter estimates of the DCBs and slant ionospheric delays are merged into combined DCB and station-dependent vertical ionospheric delays. The local station filters operate at an update rate of 1 Hz, the global clock merging filter operates at a reduced rate of once every 5 seconds and the DCB/iono-filter is updated every 60 seconds. The system is capable of processing up to 150 stations and all GPS, GLONASS, Galileo, BeiDou and QZSS satellites. The local station filters can process all signals emitted by the satellites and process single, uncombined observations. Besides the real-time data streams with GNSS observations and broadcast ephemerides, the system requires file-based input data for predicted satellite orbits, Earth-orientation parameters, and satellite and station meta-data information like antenna or receivers types and station coordinates. The presentation will provide detailed insight into the algorithms and filter design of the newest version of RETICLE for processing GPS, GLONASS, Galileo, BeiDou and QZSS. Results for multi-GNSS clock estimates will be presented.