In this study, exptl. measurements of the bubble points of binary mixtures containing varying concentrations of CO2 (99.5 %, 99 %, 98.5 %,98 %, and ∼ 95 %) with hydrogen (H2) were made.These measurements were carried out from low temperatures (240.20 K) up to 294.84 K (with uncertainties of 0.14 K) using the constant composition expansion method.The exptl. data were used to validate two thermodn. models - the Peng-Robinson and Multi-Fluid Helmholtz Energy Approximation Equation of state (PR-EoS and MFHEA-EoS).From the results, the presence of H2 in CO2 at concentrations between 0.5 and 5 % caused a significant (∼19-980 %) pos. deviation from the phase behavior of CO2 stream compared to that of pure CO2.This effect intensified with higher concentrations of H2 and decreased with rising temperaturesBoth models demonstrated good agreement with the exptl. bubble point data, exhibiting <4 % average deviation for the system.Notably, the PR-EoS outperformed the MFHEA-EoS, showing <3 % average deviation.Densities of CO2 (99.5 %) with H2 were measured at 278.14, 298.34, 323.55, and 348.40 K and pressures up to 35 MPa.While the densities of CO2 (94.99 %) with H2 were measured at 278.06, 288.13, 298.26, and 323.53 K and pressures up to 35 MPa using a vibrating tube densimeter which was calibrated using water and hydrogen.For the 99.5 % CO2 binary mixture, the average absolute relative deviations (AARD) of the model predictions were 0.09 % and 0.26 % against MFHEA and PR EoS resp.The AARD of the model predictions for the 94.99 % CO2 were 0.33 % and 1.49 % with MFHEA and PR EoS resp.Furthermore, even at low concentrations (0.5 %), the presence of H2 led to a substantial reduction (>35 %) in the d. of the mixture compared to that of pure CO2 at lower pressure conditions with this effect becoming more pronounced at higher temperatures and concentrations of H2.Both models predicted the densities of the system well (with <2 % deviations from the exptl. data), though MFHEA-Eos was more accurate with <0.4 % maximum relative deviation (MaxRD) for all the data points.