The success of the R1 logic model published by Deepseek recently and the accompanying research article surprised the world. In the R1 research article, the Chinese initiative announced that the education cost of advanced artificial intelligence, which is equivalent to OpenAI’s O1 model, is much lower and allowed everyone to control the models by making their models open source. As can be estimated, these developments caused the shares of several major technology companies to fall.
Although some time has passed since the first shock, OpenAI CEO Sam Altman finally shared a sharing about Deepseek’s excitement and said that he was excited to have a “having a new opponent”: “Deepseek’s R1 is an impressive model, especially in terms of what they can offer according to the price. Obviously we will offer much better models, and also to have a new opponent really revitalizes. We will release some new versions. “
Altman, in the continuation of his statement, “Usually, we are excited to continue to carry out our research roadmap, and we believe that more information process is more important than ever before to succeed in our mission. The world will want to use too many AIs and will be really surprised by future generation models. I look forward to bringing all of you beyond and beyond. ”
Judging by the descriptions of OpenAI CEO, it can be said that the O3 is expected to be introduced very soon. However, although Altman emphasizes that his artificial intelligence models are more advanced and will continue to develop, the use of only 3 %to 5 %of OpenAI’s resources for similar progress with chatgPT makes a big difference.
This is the reason why finished products like chatgpt O1 cannot be offered unlimited free of charge. Companies such as OpenAI must meet costs and make profits. However, this incredible success in Deepseek’s cost -reduction can reduce the development costs of artificial intelligence services and allow users to be raised to users. In addition to all these, Deepseek’s open source triggers many developers to rapidly turn to this model despite the doubt of the software developed in China.
