Totally feel this. The closed-off and high-price strategies of OpenAI and Anthropic are creating the best catch-up window for Chinese model developers. Here are three reasons why: 1) Chinese models like DeepSeek V4, the Qwen series, KimiK2, and GLM-5 have already surpassed many cutting-edge U.S. models in areas like coding capabilities, reasoning tasks, and especially in terms of open-source and cost-effectiveness. In aspects like context handling, adaptation to Chinese scenarios, and cost-performance ratio, they have clear advantages. This short-term substitutability and sense of catching up actually pose a significant threat to U.S. closed-source models. 2) OpenAI and Anthropic’s overly restrictive policies—such as strict limitations on Chinese IPs, payment pathways, and entities linked to China—are directly pushing a large number of Chinese developers, enterprises, and cost-sensitive users toward Chinese models. This will become a strong support for domestic large models to gradually overtake. 3) The current competitive logic of U.S. large model companies still revolves around massive capital expenditures, data center construction, core performance breakthroughs, and maintaining high valuation expectations. This inevitably forces them into a closed, high-cost cycle by relying on high pricing to cover their enormous investments. Meanwhile, Chinese companies are taking a path that emphasizes cost-effectiveness, open-source ecosystems, and rapid iteration. Once model capabilities start to level out, the strategic advantages of the latter in terms of real-world application scale and market positioning will become even more pronounced.