Sangfor AI Achieves Global Top Four and China’s First Place in CyberGym Assessment
1 day ago / Read about 0 minute
Author:小编   

Recently, during the international AI security capability benchmark assessment—CyberGym—Sangfor AI, which is built on a wholly domestically developed foundational model (GLM-5.2), successfully completed 1,301 out of 1,507 tasks under a fixed model configuration. Confronted with formidable security vulnerabilities such as ARVO and OSS-Fuzz, it attained an impressive success rate of 86.3%. Its overall evaluation score not only ranked among the top four worldwide but also secured the first position in China.