We present Gibbs ensemble Monte Carlo simulations of monomer-solvent and polymer-solvent mixtures with soft interaction potentials, that are used in dissipative particle dynamics simulations. From the simulated phase behavior of the monomer-solvent mixtures one can derive an effective Flory-Huggins chi -parameter as a function of the particle interaction potential. We show that this chi -parameter agrees very well with the free energy difference between a monomer surrounded by solvent particles, and a solvent particle surrounded by solvent particles. We develop a new "identity change" Monte Carlo move to equilibrate the polymer-solvent mixtures. In this move a polymer chain from one box is exchanged with an equal number of solvent particles from the other box. At realistic densities this new move offers a large computational advantage over the convential insertion method for a polymer chain using a configurational bias Monte Carlo algorithm. The new algorithm is demonstrated for polymer-solvent mixtures with a chain length of up to 150 segments. Significant differences are found between the simulated polymer-solvent phase behavior and results predicted by mean-field theory. Finally, we fit a master-equation to the simulated binodal curves at different chain lengths. This function is used to make a quantitative comparison between the simulations and experimental data for the phase equilibrium of the polystyrene-methylcyclohexane system.