[Kimi Temporarily Not Pursuing Video Models, Citing Limited Impact on Intelligence] According to monitoring by 动察 Beating, Zhou Xinyu, co-founder and head of the algorithm team at Dark Side of the Moon, stated that Kimi currently has no plans to develop video generation models. The team is presently more focused on enhancing the intelligence ceiling of their models. He believes that while video generation is practical, it offers little benefit in improving model intelligence, whether from a theoretical or practical perspective. Kimi K3 was just released. The officially highlighted key capabilities remain software engineering, knowledge work, deep reasoning, and image understanding. Kimi's multimodal approach currently leans more toward understanding and reasoning rather than video generation. [Original Article Link]