Tag
An analysis of output similarity suggests that Chinese AI frontier models like Kimi K3, DeepSeek V4, and GLM 5.2 show stylistic differences indicating meaningful independent development, challenging claims that their progress relies heavily on distillation from US models.