A new study presents a Generative Assimilation and Prediction (GAP) framework aimed at unifying data assimilation and prediction for weather and climate modeling. The article states that reliable modeling of Earth’s chaotic atmosphere is fundamental to predicting both day-to-day weather and long-term climate change, and that data assimilation combines observations with models to infer current atmospheric states while prediction projects initial state estimates forward in time until chaos inevitably blurs the view of the future. The core challenge is accurately representing and evolving the probabilities of possible atmospheric states given sparse observations, imperfect prediction models, and inconspicuous natural and anthropogenic forcings. The GAP framework is based on the Probabilistic Assimilation and Prediction (PAP) concept, illustrated on a logarithmic timeline to highlight the duality between assimilation and prediction that the authors aim to unify. During assimilation, observations progressively constrain the climatological distribution, narrowing possible states. During prediction, uncertainty evolves through distinct phases: weather forecasting up to around two weeks shows growing uncertainty until deterministic predictions lose utility; seasonal prediction up to around 1–2 years maintains predictability below climatological levels through low-frequency climate process constraints; and climate prediction beyond 1–2 years shows complete dissipation of initial condition information while external forcings modify the underlying climatological distribution. In the assimilation phase, state estimates sampled from a probability distribution conditioned on the previous state and current observations are used with a forecasting model to generate preliminary predictions, which together with observations guide subsequent probabilistic state estimation. A transition point marks the end