At the beginning of February this year, the three body computing constellation carried out a three-star collaborative on orbit intelligent processing test for water environment monitoring. Through the surface element extraction model, key water bodies were extracted in the case of icing on the water surface, and the full link capability of "satellite load work - on orbit data processing - inter satellite collaborative transmission - on orbit model calculation - task results download" was verified. On February 12th, a reporter from Science and Technology Daily learned from Zhijiang Laboratory that the Three Body Computing Constellation has achieved a breakthrough in inter satellite networking. Through in orbit collaboration, it has deployed and verified 10 artificial intelligence (AI) models and applications, exploring innovative space computing applications in scenarios such as deep space exploration, smart city construction, and natural resource exploration. It is reported that the Three Body Computing Constellation is a thousand star scale space computing infrastructure jointly built by Zhijiang Laboratory and global partners. Through "computing space, inter satellite interconnection, and model space", it promotes the paradigm change of space research and helps the innovative development of "space+artificial intelligence". On May 14, 2025, the Three Body Computing Constellation successfully launched 12 computing satellites into orbit. According to Li Chao, the Chief Technical Engineer of the Computing Constellation Department at Zhijiang Laboratory, after 9 months of in orbit testing, the first mission of the Three Body Computing Constellation has formed four core capabilities: networking, computing, model deployment, and scientific payload in orbit verification. Sending models into space is the key to promoting the application and development of artificial intelligence in space. As of now, the team has achieved in orbit deployment of 10 artificial intelligence models and applications. Among them, the 8 billion parameter space-based remote sensing model and the 8 billion parameter space-based astronomical time-domain model are the largest parameter scale models operating in orbit worldwide; Six models and algorithms, including the 150 million parameter "Fuxi" meteorological model, the 600 million parameter Qianwen big language model, and the surface element extraction model, have been deployed for in orbit updates through ground annotation. Li Chao stated that the deployed models have successfully executed multiple in orbit missions. On November 10, 2024, the team conducted a survey experiment on infrastructure such as sports venues and bridges in a 189 square kilometer urban area in northwest China. Using a space-based remote sensing model, the team automatically identified facilities such as bridges and track and field fields under heavy snow coverage conditions. The use of artificial intelligence models in space is still driving a paradigm shift in space research. Taking astronomy research as an example, two satellites in the Three Body Constellation are equipped with cosmic X-ray polarization detectors. By deploying a space-based astronomical time-domain model, rapid in orbit determination and classification of gamma ray bursts (GRBs) can be achieved. This model reduces the amount of data transmitted from a few hundred megabytes per day to tens of kilobytes, which is only one ten thousandth of the traditional observation method. The processing time is also shortened from hours to seconds, and it maintains a high event recognition accuracy of up to 99%, providing new technical support for in orbit astronomical observations. Artificial intelligence cannot be absent from space due to a lack of computing power. ”As Wang Jian, an academician of the CAE Member and director of Zhijiang Laboratory, said, this is the original intention of the laboratory to build a three body computing constellation. The deployment and application of the model is based on the breakthrough of three body computing constellation networking and computing power. The first 12 computing satellites were all equipped with onboard computing units developed by Zhijiang Laboratory, with a maximum computing power of 744TOPS per satellite, which is 74.4 trillion operations per second. After gaining computing power, the team conducted tests on satellite ground and inter satellite networking. The first mission of the three body computing constellation realized the IPlization of all loads and ground stations, breaking the barriers between satellite networks and ground Internet. The team has recently achieved in orbit chain building for 6 satellites, marking an important step towards satellite networking and interconnection. At the same time, through the space-based distributed operating system, the first mission integrates inter satellite and satellite ground computing resources, achieving unified management and scheduling of constellation tasks, computing power, storage, network and other resources. At present, the first mission of the Three Body Computing Constellation has achieved the coordinated operation of all in orbit computing nodes, with an overall in orbit computing power of 5POPS, which means 5 quadrillion operations per second. It can support up to 140 billion parameter model in orbit deployment and inference, making it the largest space computing constellation in the world in terms of computing power scale. (New Society)