Forecasts for the constraining power of 21 cm surveys using SKA-Low for EoR model inference. Two parameters, the ionization efficiency |$\xi$| and the minimum virial temperature |${\rm log}_{10}[T_{\rm vir}/{\rm K}]$|⁠, are considered here. The correlation coefficient of the two parameters, |$\rho _{\zeta {\rm log}T_{\rm vir}}$|⁠, and the determinant of the Fisher matrix, |${\rm det}[\mathcal {F}]$|⁠, are also shown for reference. The faint and bright models are as defined in Section 2.1. From left to right, each column shows the fiducial value, the forecasts for using only the spherically averaged monopole (‘mono + avg’), the spherically averaged multipoles (‘multi + avg’), the monopole in clustering wedges (‘mono + wedge’), and the multipoles in clustering wedges (‘multi + wedge’).