This August, 27-year-old Chen Yongchao took up a position as an assistant professor at Tsinghua University's School of Artificial Intelligence and founded Chaoyan Intelligence. This startup is devoted to pioneering the next generation of foundational models that can conduct independent research, engage in continuous learning, and undergo self-evolution. Despite its official establishment being slated for June 2026, the company's core team has already burgeoned to nearly 30 members in less than three months. Recently, Chaoyan Intelligence successfully secured nearly 400 million yuan in angel and angel+ round financing. These funds are earmarked primarily for foundational model and computing power development, recursive research initiatives, and team expansion.

Chen Yongchao completed his undergraduate studies at the University of Science and Technology of China and subsequently underwent joint training at Harvard University and the Massachusetts Institute of Technology. During this time, he gradually shifted his research focus to the realm of artificial intelligence. He has held positions at several prestigious research institutions. Chen believes that while traditional disciplines often encounter high marginal costs for breakthroughs, large models in their nascent stages present clear opportunities for enhancement. By late 2025 to early 2026, with the progression of model capabilities, he made the decision to forego overseas prospects and return to China to embark on his entrepreneurial journey.

Recursive Self-Improvement (RSI), the cornerstone of Chaoyan Intelligence's strategy, has emerged as a key focus in the global technology race, with several overseas teams also actively exploring similar approaches. To this end, Chaoyan Intelligence has crafted a technical closed-loop system. In this system, once the model completes research tasks, it transforms its experiences into training data to facilitate iterative upgrades. Presently, Chaoyan Intelligence's self-evolving AI system is capable of highly automated completion of the entire scientific research process. Its autonomously generated papers have achieved outstanding results in ARR reviews. Furthermore, the company has partnered with multiple institutions to introduce ASI-Bench, a benchmark designed to evaluate AI's autonomous research capabilities.

Chaoyan Intelligence has also applied this system to the research and development of AI itself, setting new benchmarks in several public tasks and engaging in collaborative research in the AI for Math direction. Looking ahead, Chaoyan Intelligence will assess whether this methodology can be extended to a broader range of research and development scenarios, with the ultimate aim of enabling AI to create incremental knowledge.