Boris Power, the Head of Applied Research at OpenAI, revealed that the organization has channeled 80% to 90% of its research endeavors into the development of forthcoming models, such as GPT-7 and GPT-8, along with even more advanced versions. Simultaneously, OpenAI is also engaged in the ongoing refinement of its existing GPT-6 series models. Despite the internal view that this partial investment in current models is "extremely short-sighted," it enables rapid rectification of product deficiencies, aids the team in gaining valuable experience, and caters to the immediate needs of users. OpenAI's product strategy entails initially constructing top-tier large-scale models and subsequently transferring their capabilities to smaller, more cost-efficient lightweight models through techniques like knowledge distillation. This approach effectively lowers the entry barrier. In practical scenarios, large and small models can work in tandem, with large models tackling intricate tasks demanding profound judgment and small models managing simple, routine tasks, thereby comprehensively reducing overall operational costs. As models continue to evolve, AI's collaborative prowess is also on the rise, transitioning from mere tools to cooperative entities capable of working in harmony. OpenAI's ultimate objective is to alleviate user burdens, enabling large models to push the boundaries of their capabilities while making small models omnipresent in daily applications.