The characteristic frequency of thermal motion is around 7E12 Hz at room temperature (300K), but from that information how can we conclude that the bonds are hard; they don't vibrate !!
Saying that bonds are hard is just an approximation. In fact, for classical molecular dynamics simulations you need to choose the time step being smaller than the fastest characteristic time in the system. Typically, bonds vibrate around 10-15 secs, i.e. femtoseconds.
The typical time step in classical MD is 1fs, but this is possible given that bonds are treated in a rigid way. Otherwise you would need smaller timesteps.
As an example, one of the most used potentials for water is TIP4P/2005. In that case you can use a timestep of 2fs, as the bonds are considered rigid.
But, the same people developed a flexible potential f-TIP4P/2005 and they used a timestep of 0.2 fs to include the bond vibrations.
Article A flexible model for water based on TIP4P/2005
The bonds use the hopping energy and this is much higher than the thermal energy. Notice that one eV is equivalent to a thermal energy of 11604.5 K !!! Thus the phonons (accoustic or optical) don't interact practically with the bond electrons.