Artificial intelligence driven scientific research (AI for Science) is an important support for accelerating the paradigm shift and capability enhancement of scientific research in China, improving China's scientific and technological innovation capabilities, and seizing the global technological high ground. Wang Jian, an academician of the CAE Member, director of Zhijiang Laboratory and founder of Alibaba Cloud, believes that AI is not a tool revolution, but a tool of a scientific revolution or a revolutionary tool of science. Monitoring the water resources of the Qinghai Tibet Plateau, carrying out ecological analysis on coral reefs, and achieving independent astronomical observation... Under the wave of AI for Science, several research institutes of the Chinese Academy of Sciences are starting an intelligent scientific research expedition from "going to heaven" to "going to the earth", and from "observing the stars" to "observing the water" by accessing the big reasoning model of Ali Tongyi Qianwen QwQ-32B. Accurately predicting solar flares is one of the most intense activity phenomena on the Sun, with major eruptions occurring every approximately 11 years. In recent years, scholars have started to use deep learning, machine learning, and other technologies to conduct research from a data-driven perspective in order to solve the mystery of solar flares. However, with the continuous accumulation of observation data and the increasing dimensionality of data features, the requirements for algorithm scale are becoming higher and higher, and researchers urgently need to effectively process massive multimodal data. The emergence of large models brings new solutions. At the National Astronomical Observatory of the Chinese Academy of Sciences, researchers use the large solar physics model "Jinwu" built by QwQ-32B to accurately predict solar flare activities. Li Yuyang, a core member of the Artificial Intelligence Group at the National Astronomical Observatory, introduced to a reporter from Science and Technology Daily that based on the Alibaba Tongyi Qianwen series model, the team trained the model through supervised learning and reinforcement learning to understand and answer solar physics questions, as well as recognize and analyze solar images. Among them, the solar flare eruption prediction task is trained and tested mainly based on publicly available data from the SDO (Solar Dynamics Observatory) satellite, data from the 35 centimeter magnetic field telescope at Huairou base, and data from the "Kuafu-1" (ASO-S) full surface vector magnetometer. The prediction accuracy has reached the forefront level in the field. The astronomical model 'Star Language 3.0' is connected to the Mini 'Sitian' telescope array of the Xinglong Observatory of the National Astronomical Observatory, which can autonomously control observations, analyze data, and recommend follow-up plans Li Yuyang introduced that "the newly upgraded 'Star Language' big model is developing towards intelligent agents, absorbing more knowledge from segmented fields, integrating existing scientific research models and algorithms, and further improving scientific research efficiency." He also mentioned that scientific research has extremely high requirements for data security, so some scientific research scenarios will deploy models locally. QwQ-32B can provide relatively lower deployment costs and meet the requirements of relevant research on model capabilities. Exploring the "water code" of the Qinghai Tibet Plateau. The Qinghai Tibet Plateau is the "Roof of the World" and "Water Tower of Asia", and is also one of the most sensitive regions in the world to climate change response. The second Qinghai Tibet scientific expedition found that the solid water on the Qinghai Tibet Plateau is rapidly melting, while the liquid water is showing an increasing trend. The uncertainty caused by climate change will pose potential risks to water and energy supply in the Qinghai Tibet Plateau. In the face of this challenge, the Qinghai Tibet Plateau Research Institute of the Chinese Academy of Sciences, together with Alibaba Cloud, created the first large-scale water energy food multimodal reasoning model focusing on climate change adaptation - "Luoshu". The coupling of water energy food refers to the complex interdependence and impact relationship between water resources, energy systems, and food production. Studying the coupling of water energy food is crucial for developing more resilient response strategies. Xia Cuihui, assistant researcher of the Qinghai Tibet Plateau Research Institute of the Chinese Academy of Sciences, said that "Luoshu" is based on the scientific model "Siyuan" developed by the Institute and trained by relying on the spatio-temporal data of the Qinghai Tibet Plateau. The output results consist of two parts: one is the runoff that directly supports the prediction of hydropower production, and the other is high-dimensional data that accurately depicts hydrological processes. But humans cannot directly understand and use this data. After integrating with Tongyi Qianwen, 'Siyuan' can achieve natural language queries and outputs, visualize high-dimensional data, and enable frontline personnel to make decisions directly based on it, "said Xia Cuihui. It is worth mentioning that "Luoshu" combined with "Siyuan" and QwQ-32B can directly infer and analyze data, and draw conclusions. For example, questions such as what to observe during the dry season and what work to do in the future to adapt to climate change can be explained through inference models to assist decision-making, "said Xia Cuihui. In this study, the team also utilized the AI computing resources, data storage, and deep learning platform provided by Alibaba Cloud to efficiently process massive data and complex computing tasks, achieve rapid experimentation and iteration of models, and greatly improve research efficiency. In the future, "Luoshu" will also be integrated with Qwen VL to achieve efficient recognition of image data, and collaborate with intelligent agents, embodied intelligent observation, and integrated dynamic data centers for air, space, and earth to jointly provide technological support for ecological protection and sustainable development of the Qinghai Tibet Plateau. (New Society)