In the competition in the field of AI, "reasoning ability" is becoming the focus of the industry. Now, from OpenAI to Google and Anthropic, major technology companies are competing to launch large models with reasoning ability. There is a widely accepted saying in the field of AI, which has further promoted the industry's emphasis on reasoning ability, that is, the law of scale. Nvidia CEO Huang Renxun said: "Various evidence shows that as our models become larger and the more training data, the efficiency, quality and performance of intelligent systems will be better." However, Apple's recently released research report "The Illusion of Thinking" has poured cold water on this trend. They used a set of logic games to test the performance of current mainstream large models on reasoning tasks. The results show that once the problem becomes complex enough, these reasoning models will "collapse" and the accuracy rate is close to zero. However, there are also voices that believe that this report may be a strategic behavior of Apple. After all, it is still behind other competitors in the large model competition.