Chinese AI labs have released multiple strong 2026 models, including Moonshot’s Kimi K3 (July), Alibaba’s Qwen3.8-Max (August), Xiaomi’s MiMo-V2.6-Pro (September), and DeepSeek V4.1-Flash updates (October), that now lead open-weight leaderboards on Artificial Analysis and excel in coding agents, long-context tasks, and price. These mixture-of-experts systems with 1M-token windows and open weights have driven sharp adoption gains, yet they trail closed US frontier models such as Claude Opus 5.5 by roughly 10–12 points on overall intelligence indices. No confirmed releases by year-end appear positioned to close the remaining gap before December 31.