Existing CO 2 emissions reported by city inventories usually lag in real-time by a year or more and are prone to large uncertainties. This study responds to the growing need for timely and precise estimation of urban CO 2 emissions to support present and future mitigation measures and policies. We focus on the Paris metropolitan area, the largest urban region in the European Union and the city with the densest atmospheric CO 2 observation network in Europe. We performed long-term atmospheric inversions to quantify the citywide CO 2 emissions, i.e., fossil fuel as well as biogenic sources and sinks, over 6 years (2016-2021) using a Bayesian inverse modeling system. Our inversion framework benefits from a novel near-real-time hourly fossil fuel CO 2 emission inventory (Origins.earth) at 1 km spatial resolution. In addition to the mid-afternoon observations, we attempt to assimilate morning CO 2 concentrations based on the ability of the Weather Research and Forecasting model with Chemistry (WRF-Chem) transport model to simulate atmospheric boundary layer dynamics constrained by observed layer heights. Our results show a long-term decreasing trend of around 2 % ± 0.6 % per year in annual CO 2 emissions over the Paris region. The impact of the COVID-19 pandemic led to a 13 % ± 1 % reduction in annual fossil fuel CO 2 emissions in 2020 with respect to 2019. Subsequently, annual emissions increased by 5.2 % ± 14.2 % from 32.6 ± 2.2 Mt CO 2 in 2020 to 34.3 ± 2.3 Mt CO 2 in 2021. Based on a combination of up-to-date inventories, high-resolution atmospheric modeling and high-precision observations, our current capacity can deliver near-real-time CO 2 emission estimates at the city scale in less than a month, and the results agree within 10 % with independent estimates from multiple city-scale inventories.