### OpenAI's CEO Reveals GPT-4 Rebuild Requires Only 5 to 10 People In the latest episode of OpenAI's podcast, CEO Sam Altman disclosed that the company can now rebuild its GPT-4 model with just 5 to 10 people, thanks to the advancements and insights gained from the development of GPT-4.5. When GPT-4 was released in March 2023, it required the efforts of hundreds of people and nearly every resource OpenAI had. The transformation to a smaller team size for GPT-4 reconstruction highlights the efficiency and progress in AI development. GPT-4.5, which was released in February 2025, is described by OpenAI as their “largest and most powerful” model to date. Altman referred to it as the "first model that feels like talking to a thinking person" on a social media platform. This model marks significant advancements in computation, data handling, and algorithms, enabling it to tackle more complex tasks with greater efficiency. Key figures in the development of GPT-4.5, such as Alex Paino and Daniel Selsam, emphasized the reduced team size due to these advancements. Paino, the pre-training machine learning lead, mentioned that GPT-4.5’s design goal was to be "10 times smarter than GPT-4." Selsam noted that having prior knowledge of what can be achieved greatly facilitates the development process, driving innovation and improvement. OpenAI's progress is further underscored by its financial strength. In March, the company completed the largest private tech funding round in history, securing $30 billion from SoftBank and another $10 billion from other investors, valuing the company at $300 billion. Tech giants like Microsoft, Amazon, Google, and Meta are also investing heavily in AI infrastructure, expected to allocate $320 billion by 2025. According to industry insiders, OpenAI's breakthroughs demonstrate its robust research and development capabilities and clear vision for the future. These advancements not only enhance model efficiency but also provide valuable insights and inspiration for other companies in the AI sector. As a leading AI research company, OpenAI's innovations are likely to further solidify its position at the forefront of the industry. ### OpenAI to Replace GPT-4 with GPT-4o in ChatGPT OpenAI plans to phase out the GPT-4 model in its popular ChatGPT platform, replacing it with the more advanced GPT-4o model on April 30, 2025. GPT-4, launched in March 2023, quickly gained global traction for its superior natural language processing capabilities, excelling in tasks such as dialogue generation, content creation, and complex multimodal tasks like understanding and generating text from images. However, as technology evolves and user demands grow, OpenAI believes it is time to retire GPT-4 and introduce GPT-4o, which offers several enhancements. GPT-4o is designed to improve semantic understanding, generate more coherent dialogues, and handle complex tasks more effectively. It also adds support for additional languages and incorporates improvements in security and privacy protection, making ChatGPT a more reliable and versatile platform. To ensure a smooth transition for users and developers, OpenAI will continue to support GPT-4 through its API for businesses and projects that rely on it. The company also pledges to maintain the basic user experience of ChatGPT, allowing users to access and utilize the latest AI dialogue technology without extensive relearning or adaptation. In the next few weeks, OpenAI will release more detailed information about GPT-4o, including performance parameters and technical documentation, to help users and developers better understand the new model's capabilities and limitations. This move is seen as a testament to OpenAI's commitment to continuous innovation and user satisfaction. Tech experts widely regard this adjustment as a sign of OpenAI's sustained leadership and adaptability in the AI sector. The company's ability to quickly develop and deploy cutting-edge models shows its responsiveness to market needs and its determination to stay ahead in the competitive landscape. ### China and the U.S. Diverge in AI Model Integration Strategies The strategies of integrating the latest artificial intelligence (AI) models into consumer technology differ significantly between China and the United States. American companies like OpenAI and Anthropic often place their most advanced models behind paywalls, requiring users to pay for access. In contrast, Chinese tech giants such as Alibaba, Baidu, and Tencent are rapidly and widely integrating their AI models into everyday products and services for free. Washington D.C. analyst Ray Wang noted that China is embedding AI into every nook and cranny, focusing on quick deployment and practical applications rather than outpacing Western AI leaders in model development. Even though China's large language models (LLMs) may not be as advanced as those from the U.S., the rapid integration and widespread use in consumer devices and applications could provide a significant real-world advantage. In recent months, major Chinese companies have launched powerful AI models and enhanced versions. Alibaba introduced a new AI model for cost-effective AI agent development, and DeepSeek released an improved version of its open-source V3 LLM. Both the Qwen2.5-Omni-7B by Alibaba and DeepSeek's V3 model are freely available for users to download, modify, and integrate. DeepSeek plans to roll out inference-specific R1 and R2 models in the coming weeks, strengthening its market position. Tencent has deployed its Hunyuan model and DeepSeek’s R1 model across its extensive ecosystem, including the WeChat application, which has nearly 1.4 billion users. Baidu, too, has integrated DeepSeek’s R1 model into its search engine and launched new versions of its AI models, Ernie X1 and Ernie 4.5, with plans to integrate them further into its product lineup. Despite the U.S. maintaining an overall advantage in AI, with private investments reaching $109.1 billion in 2022 compared to China's $9.3 billion, the quality of Chinese AI models is rapidly improving. Chinese companies lead in AI publications and patents, and the trend of quick and widespread integration is raising their profile globally. Stanford University’s AI Index Report highlights that China, initially caught off guard by Western products like ChatGPT, is now directly competing with top Western suppliers. While China lags in AI chip manufacturing, it has made significant strides in software and applications, providing viable alternatives to users seeking non-U.S. AI solutions. Lian Jye Su, an industry expert, believes that China’s rapid integration strategy has yielded notable success. Tech companies can swiftly adapt and develop competitive AI products, enhancing user acceptance and opening up more market opportunities. This strategic shift may give Chinese tech firms a new edge in the global AI race. Industry insiders have praised OpenAI's continuous advancements, noting that the company's ability to innovate and scale efficiently is setting new benchmarks in the AI sector. OpenAI’s steady improvements and aggressive funding, coupled with the global investments in AI infrastructure, are shaping the future of AI technology, driving both scientific and commercial progress.