jats:titleAjats:scbstract</jats:sc> </jats:title>jats:pWe juxtapose global fits of two bottom-up models (an jats:italicS</jats:italic>jats:sub3</jats:sub> scalar leptoquark model and a jats:italicB</jats:italic>jats:sub3</jats:sub> – jats:italicL</jats:italic>jats:sub2</jats:sub>jats:italicZ′</jats:italic> model) of jats:italicb → sμ</jats:italic>jats:sup+</jats:sup>jats:italicμ</jats:italic>jats:supjats:italic−</jats:italic></jats:sup> anomalies to flavour data in order to quantify statistical preference or lack thereof. The leptoquark model couples directly to left-handed di-muon pairs, whereas the jats:italicZ′</jats:italic> model couples to di-muon pairs with a vector-like coupling. jats:inline-formulajats:alternativesjats:tex-math$$ {B}s-\overline{B_s} $$</jats:tex-math><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:msub mml:miB</mml:mi> mml:mis</mml:mi> </mml:msub> mml:mo−</mml:mo> mml:mover mml:msub mml:miB</mml:mi> mml:mis</mml:mi> </mml:msub> mml:mo¯</mml:mo> </mml:mover> </mml:math></jats:alternatives></jats:inline-formula> mixing is a focus because it is typically expected to disfavour jats:italicZ′</jats:italic> explanations. In two-parameter fits to 247 flavour observables, including jats:italicB</jats:italic>jats:subjats:italics</jats:italic>/jats:italicd</jats:italic></jats:sub>jats:italic→ μ</jats:italic>jats:sup+</jats:sup>jats:italicμ</jats:italic>jats:supjats:italic−</jats:italic></jats:sup> branching ratios for which we provide an updated combination and LHCb jats:inline-formulajats:alternativesjats:tex-math$$ {R}{K^{\left(\ast \right)}} $$</jats:tex-math><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:msub mml:miR</mml:mi> mml:msup mml:miK</mml:mi> mml:mfenced mml:mo∗</mml:mo> </mml:mfenced> </mml:msup> </mml:msub> </mml:math></jats:alternatives></jats:inline-formula> measurements from December 2022, we show that each model provides a similar improvement in quality-of-fit of jats:inline-formulajats:alternativesjats:tex-math$$ \sqrt{\Delta {\chi}^2} $$</jats:tex-math><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:msqrt mml:mrow mml:mtext∆</mml:mtext> mml:msup mml:miχ</mml:mi> mml:mn2</mml:mn> </mml:msup> </mml:mrow> </mml:msqrt> </mml:math></jats:alternatives></jats:inline-formula> = 3jats:italic.</jats:italic>6 with respect to the Standard Model. The main effect of the jats:inline-formulajats:alternativesjats:tex-math$$ {B}_s-\overline{B_s} $$</jats:tex-math><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:msub mml:miB</mml:mi> mml:mis</mml:mi> </mml:msub> mml:mo−</mml:mo> mml:mover mml:msub mml:miB</mml:mi> mml:mis</mml:mi> </mml:msub> mml:mo¯</mml:mo> </mml:mover> </mml:math></jats:alternatives></jats:inline-formula> mixing constraint in the jats:italicZ′</jats:italic> model is to disfavour values of the jats:italics</jats:italic>jats:subjats:italicL</jats:italic></jats:sub>jats:italic– b</jats:italic>jats:subjats:italicL</jats:italic></jats:sub> mixing angle greater than about 5|jats:italicV</jats:italic>jats:subjats:italiccb</jats:italic></jats:sub>|. This limit is rather loose, meaning that a good fit to data does not require ‘alignment’ in either quark Yukawa matrix. No curtailment of the jats:italics</jats:italic>jats:subjats:italicL</jats:italic></jats:sub>jats:italic− b</jats:italic>jats:subjats:italicL</jats:italic></jats:sub> mixing angle is evident in the jats:italicS</jats:italic>jats:sub3</jats:sub> model.</jats:p>