Photo/VCG

On the evening of April 11th, a journalist from NBD learned that during a recent internal speech, Robin Li shared his insights on the industry's hot topics, such as the choice between open-source and closed-source models, and whether AI entrepreneurs should focus on models or applications.

During the internal speech, Robin Li mentioned that closed-source models will continue to lead in capabilities, not just temporarily; open-sourcing models is not a case of "the more people involved, the higher the flames." This is quite different from traditional software open-source projects, such as Linux and Android.

Robin Li also stated that closed-source models have a business model and are profitable, which allows them to attract computing power and talent. Closed-source models are actually cost-effective; The inference cost of closed-source models will always be lower, and the response speed will always be faster.

Furthermore, Robin Li pointed out that the strongest foundational models in both China and the United States are closed-source. Models derived from these foundational models through dimensionality reduction are also better, giving closed-source models an advantage in cost and efficiency. For AI entrepreneurs, the core competitiveness should not be the model itself, as it is too resource-intensive and requires long-term commitment to succeed.

Robin Li believes that the "dual-wheel drive" approach of doing both model development and application is not a good model for startups. Startups have limited resources and energy, and they should focus more. Splitting efforts between model development and application will inevitably dilute their focus.