<p>The Global Carbon Project (GCP) compiles an updated global carbon budget each year, synthesizing state‑of‑the‑art estimates of anthropogenic CO₂ emissions, land and ocean sinks, and the atmospheric CO₂ growth rate. The residual between these terms, referred to as the global carbon budget imbalance, reflects the aggregate inaccuracies of the individual component estimates. Growth rates derived from marine boundary layer (MBL) surface flask mixing ratio observations are assumed to be highly accurate. Hence, land and ocean sink estimates from process models are viewed as the primary source of the imbalance. Here we show that substantial discrepancies arise when marine boundary layer growth rate estimates are used to represent the whole atmosphere. Correcting for this discrepancy using atmospheric flux inversion estimates reduces the 0.76 petagrams of carbon per year (PgC yr⁻¹) root-mean-square (RMS) imbalance (from the 2023 GCP report) by up to 25%. Further investigation into the imbalance metric between the 2017 and 2023 GCP reports shows a reduction in imbalance resulting from updates to each carbon budget component, leading to a 16% overall reduction. These reductions provide quantitative evidence of improvements in process models and inventory emission estimates, driven by enhanced forcing data and the inclusion of new carbon cycle processes. Overall, we report a 37% reduction in the root-mean-square imbalance, from 0.91 to 0.57 PgC yr⁻¹, between the 2017 and 2023 GCP reports by combining process model and inventory improvements with atmospheric growth rate corrections. Our findings indicate that land and ocean process models are more accurate than previously believed and that the scientific understanding of Earth’s carbon cycle is improving.</p>