WeChat article claims model 'comprehensively surpasses' Opus 4.8; author questions claim

The author encountered a WeChat article claiming a comprehensive surpass of Opus 4.8. They note that on benchmarks with definite answers and RL/verification (math and programming) it is possible to score above Opus 4.8 and GPT-5.6, but they question whether a claim of 'comprehensive' surpass is realistic or exaggerated.