Huaxiu Yao Notes Qwen3.8 Max-0902 RSI-Exam Gain
Assistant professor highlights Alibaba Qwen model's score jump in recursive self-improvement benchmark.
TLDR
Huaxiu Yao, a tenure-track assistant professor at UNC Chapel Hill who leads the AIMING Lab, posted that Qwen3.8 Max-0902 by Alibaba_Qwen raised its RSI-Exam 0.1 score from 0.322 to 0.392. The post calls the change a nearly 22 percent improvement and includes a screenshot of the leaderboard. The message frames the update as stronger performance on recursive self-improvement tasks. No independent verification appears in the supplied lines.
Combined views
61K
4 Sources, first seen 27d ago