A key objective of the ESA Gaia satellite is the realization of a quasi-inertial reference frame at visual wavelengths by means of global astrometric techniques. This requires an accurate mathematical and numerical modeling of relativistic light propagation, as well as double-blind-like procedures for the internal validation of the results, before they are released to the scientific community at large. Aim of this work is to specialize the Time Transfer Functions (TTF) formalism to the case of the Gaia observer and prove its applicability to the task of Global Sphere Reconstruction (GSR), in anticipation of its inclusion in the GSR system, already featuring the suite of RAMOD models, as an additional semi-external validation of the forthcoming Gaia baseline astrometric solutions. We extend the current GSR framework and software infrastructure (GSR2) to include TTF relativistic observation equations compatible with Gaias operations. We use simulated data generated by the Gaia Data Reduction and Analysis Consortium (DPAC) to obtain different least-squares estimations of the full stellar spheres and gauge results. These are compared to analogous solutions obtained with the current RAMOD model in GSR2 and to the catalog generated with GREM, the model baselined for Gaia and used to generate the DPAC synthetic data. Linearized least-squares TTF solutions are based on spheres of about 132,000 primary stars uniformly distributed on the sky and simulated observations spanning the entire 5-yr range of Gaias nominal operational lifetime. The statistical properties of the results compare well with those of GREM. Finally, comparisons to RAMOD@GSR2 solutions confirmed the known lower accuracy of that model and allowed us to establish firm limits on the quality of the linearization point outside of which an iteration for non-linearity is required for its proper convergence.