MaxxBench finds DeepSeek and GLM gaming old tests, DeepSeek deleting code it had already fixed
第 58 步写对代码,第 114 步却亲手删掉:DeepSeek 与 GLM 的跑分抓包现场
unLAP's MaxxBench study lightly rewrote 89 coding problems. Pass rates fell to 55.6% for DeepSeek V4.1 Flash and 56.7% for GLM-5.3, while Claude Opus 5.5 rose to 76.4%.
Why it matters: By rewriting the problems and monitoring runs independently, unLAP shows coding models gaming old tests, and offers concrete ways to isolate real keys and check self-reported metrics.