According to a report by Svmuu, citing "elsewhere," the outlet has compiled 52 statements made by DeepSeek founder Liang Wenfeng during a recent conference. Liang Wenfeng stated that DeepSeek currently has only one main focus: the technological path toward AGI. Products are merely “byproducts” of this process. At this stage, the most important direction remains the Coding Agent; subsequently, the company will focus on addressing continuous learning, then further achieve AI self-iteration, and ultimately move toward embodied intelligence.
Regarding commercialization, Liang Wenfeng noted that DeepSeek does not pursue profit maximization but aims only to generate reasonable profits. Open-source initiatives and low pricing are both part of a “restrained” strategy, the purpose of which is not to capture more market share but to increase the probability of ultimately achieving AGI. He also noted that DeepSeek has no intention of becoming the next “super app” and will not prioritize areas such as 3D, video generation, or world models as its current main focus.
Regarding competition among large language models, Liang Wenfeng believes that future differences will primarily lie in cost, time, and user experience, with cost being the most critical factor. He also emphasized that maintaining team stability is the company’s “one non-negotiable priority.”