On May 28, 2026, SuperCLUE unveiled its most recent assessment outcomes for Chinese large language models, encompassing 21 widely-used models. The evaluation framework comprised six primary tasks: mathematical reasoning, scientific deduction, code creation, agent (task planning), precise instruction adherence, and hallucination mitigation, featuring a total of 492 questions. The findings revealed that the global top tier was predominantly occupied by four overseas models: Gemini, GPT-5.5, Claude-Opus, and Gemini-Flash. The leading three Chinese models—DeepSeek-V4-Pro, Qwen3.7-Max, and Doubao Seed 2.0 Pro—achieved closely matched scores, with their global standings clustered around the fifth position, thus constituting the domestic top tier. Although there remains a disparity when compared to the international elite, Chinese models have demonstrated remarkable advancements, frequently securing top global rankings in domains such as code generation and mathematical reasoning, while also boasting exceptional cost-efficiency.