Two families of multi-planet systems with distinct architectures and stellar hosts

Abstract The rapidly growing census of exoplanets now includes over a thousand multi-planet systems, revealing architectures that span more than four orders of magnitude in mass and five in orbital scale. Explaining the origin of this striking diversity remains a central challenge for planet formation theory, compounded by observational biases that limit access to Solar System-like regimes. Here we introduce architecture-agnostic metrics to uniformly analyse four independently defined samples of multi-planet systems, uncovering unifying patterns across their diversity. We identify a clear bimodality in system masses, dividing known architectures into two distinct families. The lower-mass family, predominantly hosted by subsolar-to-solar stars, exhibits tightly packed configurations with mass distributions consistent with the ‘peas-in-a-pod’ pattern and modal dynamical excitation comparable with that of the Solar System. By contrast, the high-mass family forms in metal-rich environments around solar-to-supersolar stars, displays gravitationally compact architectures and is an order of magnitude dynamically hotter than both the Solar System and the lower-mass counterparts. Their high level of dynamical excitation points to violent evolutionary pathways driven by global instabilities and planet–planet scattering. Together, these results show that the architectures of planetary systems are born from the combined influence of formation processes and host star properties. Signatures of these roots are preserved across the post-formation dynamical evolution that reshaped most known multi-planet systems.