There is no such cutoff. Also, the distinction between covalent and reversible inhibition is not accurate. There are many examples of reversible covalent inhibitors.

There are noncovalent inhibitors that are in rapid equilibrium with the target. Their potency can be characterized by a dissociation constant, Ki (or Kd). Their potency is not time-dependent (except on a seconds-to-minutes time scale).

There are noncovalent inhibitors that are slow-binding, exhibiting a gradual increase in potency due to slow conformational changes in the target, but are in equilibrium with the target. They transition from a relatively high Ki to a relatively low Ki on the seconds-to-minutes time scale. The rate of dissociation of the tightly bound form, may be very slow.

There are covalent inhibitors that react with the target irreversibly. Their potency also increases with time, but they can't be characterized just by a Ki because they are not in equilibrium. Instead, they are characterized by kinact, the rate constant for the covalent reaction, Ki for the initial binding interaction (which may be difficult to measure because of the ongoing covalent reaction), and the bimolecular rate constant kinact/Ki.

There are covalent inhibitors that react with the target reversibly. They are also characterized by kinact/Ki, bit also by the rate constant for dissociation koff. Some examples of these are boronic acids, oxaboroles and diazabicyclooctanes.

All of these kinds of inhibitors can have rate constants and equilibrium constants with any value.

IC50 can be measured for any type of inhibitor, but its definition must be stated clearly because it is dependent on conditions and, in some cases, time. It is useful for comparing the potencies of compounds to each other under the same assay condition, but is not a replacement for actual rate constants or equilibrium constants.

Dear Adam B Shapiro , Thanks so much for your kind reply. It really resolved my long-standing confusion.