Jointly organised by the Behave Lab, University of Milan, and University of Brescia, the Behave school aims to train students in agent-based modelling through a combination of theoretical and practical lectures, daily hands-on sessions, project presentations, and consultations with experts. The school is structured in two weeks, the first introductory (31 August-4 September 2026), the second advanced (7-11 September 2026), has a hybrid setting, with up to 40 participants present and an unlimited number of others connected remotely. Training takes place from 9:30–13:00 and 14:00–17:00 CET. All lectures are streamed online, with interaction between students and lecturers, both in person and remotely. All lectures can also be accessed offline during and after the course. Students are supported by assistants in person and online. The first week is intended for beginners or experts requiring a refresher course, and will introduce the principles and methods of agent-based modelling and practical modelling exercises. The second week provides advanced training in model calibration and validation, with in-depth training in statistical techniques for parameter estimation and analysis. During both weeks, classic and contemporary agent-based models will be coded from scratch in class, step by step, with faculty members providing one-to-one assistance to each student. Students can decide whether to attend only the intro training, only the advanced training or both, either onsite or online or a mix. For in presence students, the fees include accommodation in a private room in a residence together with the faculty.