Video | OpenAI Chairman Discusses Kimi K3: Open Source Models Not Necessarily More Cost-Effective

On July 20, Bret Taylor, Chairman of the OpenAI Board, was interviewed by CNBC. When asked about his views on China's newly launched open-source model Kimi K3, he mentioned that when companies choose AI models, what truly matters is not whether it is "open source" or "closed sources," but whether the token cost investment can bring sufficient actual value.

He pointed out that open-source models are not necessarily cheaper because they may require more tokens to complete the same task. On the other hand, cutting-edge models like OpenAI and Anthropic have higher individual token prices, but they also have higher efficiency and capability. Therefore, to determine whether a model is cost-effective, one should not only look at the token price but also consider the total cost of completing the task and the final outcome.