Aims. Our long-term aim is to derive model-independent stellar masses and distances for long period massive binaries by combining apparent astrometric orbit with double-lined radial velocity amplitudes (SB2).Methods. We followed-up ten O+O binaries with AMBER, PIONIER and GRAVITY at the VLTI. Here, we report on 130 astrometric observations over the last seven years. We combined this dataset with distance estimates to compute the total mass of the systems. We also computed preliminary individual component masses for the five systems with available SB2 radial velocities.Results. Nine of the ten binaries have their three-dimensional orbit well constrained. Four of them are known to be colliding wind, non-thermal radio emitters, and thus constitute valuable targets for future high angular resolution radio imaging. Two binaries break the correlation between period and eccentricity tentatively observed in previous studies. This suggests either that massive star formation produces a wide range of systems, or that several binary formation mechanisms are at play. Finally, we found that the use of existing SB2 radial velocity amplitudes can lead to unrealistic masses and distances.Conclusions. If not understood, the biases in radial velocity amplitudes will represent an intrinsic limitation for estimating dynamical masses from SB2+interferometry or SB2+Gaia. Nevertheless, our results can be combined with future Gaia astrometry to measure the dynamical masses and distances of the individual components with an accuracy of 5 to 15%, completely independently of the radial velocities.