Subseasonal-to-seasonal (S2S) prediction remains a key challenge in extended range weather prediction, requiring improved representation of processes that govern predictability across interacting components of the Earth system. Sources of skill at these timescales, including the Madden-Julian Oscillation (MJO), El Niño-Southern Oscillation (ENSO), land-atmosphere coupling, and stratosphere-troposphere interactions, are often limited by systematic model biases, initialization challenges, and incomplete process representation. This session invites contributions that advance understanding of S2S predictability and improve model skill by identifying, diagnosing, and addressing model errors across components of the Earth system. We welcome studies including the diagnosis of model errors and bias, development of process-based diagnostics and model evaluation frameworks, advances in physical parameterizations, and improvements in initialization and data assimilation. The session will also include highlights from the cross-agency August 2026 Land-Atmosphere S2S Workshop, with a focus on strengthening coordination across modeling, observational, and application communities to advance forecast performance and confidence.