See, e.g. reaction 12, in GRI 3. b is zero, but KineticsModel is still being set as Kooij for each of the forward rates. This is minor, but it is inconsistent with us taking elementary or three-body reactions and doing to conversion internally (to save some cpu cycles).