How robust are the parameter constraints extending the ΛCDM model?

We present model-marginalized limits on the six standard $\mathrm{\ensuremath{\Lambda}}\mathrm{CDM}$ cosmological parameters (${\mathrm{\ensuremath{\Omega}}}_{\mathrm{c}}{h}^{2}$, ${\mathrm{\ensuremath{\Omega}}}_{\mathrm{b}}{h}^{2}$, ${\ensuremath{\theta}}_{\mathrm{MC}}$, ${\ensuremath{\tau}}_{\mathrm{reio}}$, ${n}_{s}$ and ${A}_{s}$), as well as on selected derived quantities (${H}_{0}$, ${\mathrm{\ensuremath{\Omega}}}_{\mathrm{m}}$, ${\ensuremath{\sigma}}_{8}$, ${S}_{8}$ and ${r}_{\mathrm{drag}}$), obtained by considering several extensions of the $\mathrm{\ensuremath{\Lambda}}\mathrm{CDM}$ model and three independent cosmic microwave background (CMB) experiments: the Planck satellite, the Atacama Cosmology Telescope, and South Pole Telescope. We also consider low redshift observations in the form of baryon acoustic oscillation (BAO) data from the SDSS-IV eBOSS survey and supernovae (SN) distance moduli measurements from the Pantheon-Plus catalog. The marginalized errors are stable against the different minimal extensions of the $\mathrm{\ensuremath{\Lambda}}\mathrm{CDM}$ model explored in this study. The largest impact on the parameter accuracy is produced by varying the effective number of relativistic degrees of freedom (${N}_{\mathrm{eff}}$) or the lensing amplitude (${A}_{\mathrm{lens}}$). Nevertheless, the marginalized errors on some derived parameters such as ${H}_{0}$ or ${\mathrm{\ensuremath{\Omega}}}_{\mathrm{m}}$ can be up to 2 orders of magnitude larger than in the canonical $\mathrm{\ensuremath{\Lambda}}\mathrm{CDM}$ scenario when considering only CMB data. In these cases, low redshift measurements are crucial for restoring the stability of the marginalized cosmological errors computed here. Overall, our results underscore remarkable stability in the mean values and precision of the main cosmological parameters once both high and low redshift probes are fully accounted for. The marginalized values can be used in numerical analyses due to their robustness and slightly larger errors, providing a more realistic and conservative approach.