From April 10 to 13, 2026, The 2nd International Geo-Energy Frontier Forum was held in Zhengzhou, with a parallel session for geo-energy academic journals held concurrently, focusing on cutting-edge fields such as green development of shale oil and gas, and CO₂ geological utilization and storage. Yang Qin, Professor at Beihang University and Chairman of GridWorld, was invited to deliver a keynote report entitledResearch Progress of Automatic Geological Modeling Technology Based on tEgg Generative Pre-trained AI Large Geological Model, sharing innovative achievements and industrial practices of AI-driven geological modeling.

Professor Yang Qin pointed out that traditional 3D geological modeling has long been plagued by pain points including data scarcity, shortage of experts, high barriers to entry and long cycles, failing to cover most long-tail scenarios and restricting the digital and intelligent transformation of the industry. With years of technological accumulation, GridWorld has relied on core products such as ShenTan Geoscience Modeling Software and Transparent Earth Engine to tackle key technical challenges including completeness, continuous evolution, integrity, meso-scale and discretization, aiming at the construction, sharing and services of global 3D geological models.

The report highlighted GridWorld’s self-developed tEgg generative pre-trained AI large geological model. Adopting a dedicated neural network architecture and pre-trained on massive calibrated data, the model can fully automatically generate high-precision 3D geological models with limited data or even no supplementary data, effectively breaking through the bottlenecks of dependence on data and experts. The on-site demonstration of modeling results with faults and under multiple constraints verified the model’s stable generation capability in complex structural scenarios.

In addition, the tEgg generative pre-trained AI large geological model can provide efficient solutions for scenarios such as resource exploration, disaster prevention and control, and engineering construction. At present, four business models have been formed, including integrated modeling and reasoning machine, MaaS cloud service, privatized customization and global geological model information service, fully meeting the needs of different customers.

Professor Yang Qin stated that in the future, the team will continue to improve the product system, enhance the model’s generalization ability, expand model types from structure to attributes and from geology to mechanics, promote multimodal joint training and language-controlled generation, and achieve the leap from automatic modeling to intelligent interpretation. Meanwhile, in response to the demands for data, computing power and professional knowledge, he called on all sectors of industry, university and research to deepen cooperation to jointly promote the innovation and application of AI geological technologies.