The 'Hundred Model Battle' has arrived. Since the beginning of this year, AI (Artificial Intelligence) large model products have emerged one after another. On May 6th, iFlytek released the Spark Cognitive Model and its application results in industries such as education, office, automotive, and digital employees. Previously, Baidu took the lead in releasing ERNIE Bot, and Ali followed closely with the official announcement of the Tongyi Qianwen, discussing that technology companies such as technology, cloud Zhisheng, and the Medical Association also announced or would soon release their AI model products. Tsinghua University, Beijing Zhiyuan Artificial Intelligence Research Institute, Shanghai Artificial Intelligence Laboratory and other scientific research institutions also released their AI model results. Prior to this, the Political Bureau of the Central Committee of the Communist Party of China held a meeting on April 28th to analyze and study the current economic situation and work, proposing to attach importance to the development of universal artificial intelligence, create an innovative ecosystem, and pay attention to risk prevention. According to incomplete statistics, there are currently over 30 AI big models released by domestic technology giants, startups, and research institutes, and many are still under development. It is controversial that the AI pre training large model, which has been developed in just a few months, is growing in sync with the quantity, and many industry insiders are concerned about its true capabilities? In the current era of increasingly fierce competition in the next generation of artificial intelligence, should the AI big model field be "blooming" or "concentrated"? How to break through technological bottlenecks such as computing power and data? How can China leverage its comparative advantage in the face of international competition in the field of AI big models? What is the real ability of the large model? Industry insiders: There is an urgent need for evaluation. Although many companies have explored pre trained AI models before, this round of "Hundred Model Battle" only reached a climax after OpenAI released ChatGPT. In March, Baidu released the big language model "ERNIE Bot"; In April, Alibaba's "Tongyi Qianwen" began enterprise internal testing; Shangtang Technology has brought "SenseChat for Discussion", and 360 has also launched the "360 Intelligent Brain" enterprise internal testing. In addition, Tencent, Jingdong, ByteDance and other domestic technology "giants" have also announced their layout in the field of ChatGPT like products or AI big models. According to statistics from research reports related to Minsheng Securities, at least 30 large models have appeared in China, including internet giants, AI concept listed companies, server leading enterprises, research institutes, and primary market startup companies. Some large models have parameter scales that are close to or even exceed the scale of ChatGPT (hundreds of billions). Based on these large models, ChatGPT like products are launched, with main capabilities including dialogue Q&A, text and code generation, etc. Despite having so many AI big model products, both professionals and ordinary users still have a misconception: how to judge the true capabilities of AI big models? With so many AI big models and GPT like products, who has the real strength? On May 6th, Liu Qingfeng, the chairman of iFlytek at the University of Science and Technology, said in an interview with reporters, "For the general model, with so many companies in China doing it, we need a relatively objective and systematic evaluation method. We need to learn from OpenAI, and also quickly catch up and strive to surpass it, which requires a scientific and systematic evaluation system