Nitrogen plays a major role in agroecosystems as the key nutrient in agricultural production as well as a source of different emissions, which exceed currently planetary boundaries. N losses, conditioned by both pedoclimatic conditions and agricultural strategies (e.g. rotations, fertilisation), predominantly take the form of ammonia (NH3) volatilisation, nitrate (NO3) leaching, nitrification-driven nitric oxide (NOx) emission to air, and denitrification-driven nitrous oxide (NOx and N2O) emissions to air. The multiplication of initiatives and studies on nitrogen modelling resulted in a broad offer of complex simulation models (Tier 3) on one extreme of the gradient between feasibility and integration of processes. On the other side, a multiplication of initiatives has led to a broad offer of causal indicators in the form of proxies and considering one or a few input variables (Tier 1). A relevant compromise between those extremes lies in the development of operational models using a restricted number of parameters and input variables (Tier 2). Here, we propose a new semi-mechanistic operational model for the estimation of direct field N emissions (NH3, NO3, NOx and N2O) from contrasting agricultural situations: the Indigo-N v3 (I-N3) model. The gaseous emissions are based on Tier 1 (NOx) and Tier 2 (NH3, N2O) methods taken from the literature, with some enhancements, while we developed a totally new semi-mechanistic approach for nitrate leaching. A comparison of I-N3 outputs was performed with measurements of nitrate leaching in three countries (15 arable fields in France, 3 sugar cane fields at Reunion Inland, and 5 cropped fields in Kenya) and showed a reasonable predictive quality for temperate arable fields, and for some of the tropical fields (1 in Reunion and 3 in Kenya). It also performed better than the previous version of Indigo-N (IN-2) and the SALCA/SQCB models. In comparison with previous Tier 2 models, the newly developed Indigo-N v3 presents an origin