@article{Zhang2026, 
author = {Guohua Zhang and Chong Wang and Wei Dong and Qianchuan Zhao},
title = {C2 multi-agent factory: large-small model collaboration empowering a new command and control paradigm},
year = {2026},
journal = {C2 Science and Technology},
keywords = {command and control system, large-small model collaboration, single-agent factory, C2 multi-agent factory},
url = {https://www.sciopen.com/article/10.26599/C2ST.2026.9700003},
doi = {10.26599/C2ST.2026.9700003},
abstract = {Command and control (C2) is an agile crisis management architecture for highly uncertain emergencies, originating from military operations but now widely deployed in public safety emergency response, traffic and airspace management, industrial energy systems, urban governance, and large-scale event coordination. Traditional C2 systems mainly rely on preset rules, static processes and manual decision-making, which expose obvious limitations in the face of modern environments with high dynamics, complex disturbances and incomplete information. Aiming at the significant problems of traditional command and control systems, such as structural rigidity, information overload, decision lag, and insufficient collaboration, based on the concepts of software factory and single-agent factory, combined with the latest progress of large-small model collaboration in the field of command and control, this paper proposes a new paradigm of large-small model collaboration empowering command and control—the concept of C2 multi-agent factory, which upgrades the single-agent factory that produces “individual intelligence” to a multi-agent factory that produces and runs “swarm intelligence”, and is combined with the command and control system and its application scenarios. The overall system architecture of the C2 multi-agent factory and the large-small model collaboration mechanism based on the C2 multi-agent factory are given. Finally, an application example is given to further illustrate how the C2 multi-agent factory can use the large-small model collaboration technology to empower command and control. The C2 multi-agent factory proposed in this paper will find a new path for the construction of an evolvable intelligent command and control system, promote the substantive leap of artificial intelligence towards the C2 system.}
}