Open models recap: more on Kimi K3, Qwen 3.8, Xi's WAIC speech, distillation, the open-closed gap, and what's next

The global AI landscape is undergoing a significant shift as Chinese labs, including Moonshot AI and Zhipu AI, rapidly narrow the performance gap with closed-source frontier models. These organizations leverage capital efficiency and specialized talent to produce competitive models like Kimi K3 and GLM, challenging the dominance of US-based providers. Contrary to popular narratives, distillation is not the primary mechanism driving these advancements; while useful for supervised fine-tuning, frontier-level reasoning increasingly depends on large-scale reinforcement learning. As near-frontier capabilities become commoditized, the industry faces structural risks, particularly regarding cybersecurity, where restricted access to top-tier models may leave domestic defenders at a disadvantage compared to global attackers. Future progress will likely see a continued rise in open-weight model adoption, with US-based players needing to innovate on model size and fine-tunability to maintain relevance in this rapidly evolving ecosystem.