The 12-day OpenAI conference has sparked industry attention but is criticized for lacking significant highlights and falling short of expectations. This conference covered multiple important updates in technology, product form, business model, and industrial ecology. However, GPT-5 was not released. The o3 released on the last day is the next-generation reasoning model of o1. It has astonishing performance in multiple tests and is considered as “AGI has arrived.” But there is still a gap between powerful models and practical applications. Here are several key points:
**1. The depth of intelligence of o3 is sufficient, but whether it can be called AGI depends on the breadth of intelligence.**
o3 outperforms some human experts in complex issues such as mathematics and coding. The technical paradigm of the large model wave has shifted from the pre-training Scaling Law to the reasoning Scaling Law. o3 is moving towards a “supercomputer with extremely high intelligence.” But only when AGI is popularized and made accessible to all will it mean a real intelligent revolution. For domestic large model companies, the core question is how to catch up with o3. Some model leaders believe that an open-source model at the o3 level may emerge in a year.
**2. The model is just an engine; the key is to help developers put it to use.**
Although the o3 model is powerful, it is still far from practical applications. Currently, the main constraints are data and cost. OpenAI has released functions such as AI-enhanced fine-tuning to help models be applied in practice, giving developers confidence in application business in the new year.
**3. Sora’s video generation is below expectations, but product openness will enhance its physical simulation ability.**
At the beginning of the year, OpenAI’s release of Sora’s demo caused a sensation, but this official release is below expectations. Sora provides some editing functions to enhance the video creation experience. Currently, domestic text-to-video companies are advancing in parallel with OpenAI. After Sora is released, its physical simulation ability may be enhanced as more people try it.
**4. With internal added functions and external connected ecology, can ChatGPT become a Super App?**
At the conference, OpenAI on the one hand added new product functions and optimized user experience, transforming ChatGPT into a super AI assistant. On the other hand, it actively expanded its ecology and integrated into terminal devices and operating systems to reach a wider audience. However, some people are not optimistic about its continuous addition of functions. Although o3 has demonstrated technical strength, there are still doubts about OpenAI’s technological development. Its focus on product form and practical construction may be a way of thinking.
