User blog comment:PsiCubed2/Question: General breakdown of levels up to Second Order Arithmetic/@comment-30118230-20180204192313/@comment-30118230-20180206204746

An axiom,stating that for every $$\Pi^1_3$$ formula with parameters holds in a beta-model of $$\Delta^1_2-CA$$.

In nLab (the link you posted above) this is the definition stated for (M),but I believe it should be $$\Pi^1_2$$ instead of $$\Pi^1_3$$ if I'm not mistaken.