Published • loading... • Updated
China’s Z.ai Unveils GLM-5.3, Claims Chart-Leading Scores
Z.ai said GLM-5.3 scored 84.5% on CyberGym and will add safeguards before a public release in about two weeks.
- On Friday, Beijing-based startup Z.ai announced its open-source GLM-5.3 model scored 84.5% on CyberGym, nearing Anthropic's restricted Mythos 5 in identifying software vulnerabilities and bolstering Chinese Artificial Inteligence credentials among Western developers.
- The launch builds on global interest in the firm's previous GLM model, which gained attention in recent months among overseas developers for coding capabilities that users said approached leading American models but at much lower cost.
- In separate timed tests, GLM-5.3 completed 105 attack-development tasks in two hours and 130 in six hours, while scoring 54.4% on ExploitBench versus 78.0% for Mythos 5, and boasting a 50pc improvement from the older version in internal benchmarks.
- Z.ai will release GLM-5.3 publicly in about two weeks after completing security assessments. Gabriel Wagner, an Artificial Inteligence governance researcher at Beijing-based Concordia, called this "the first time a Chinese lab is publicly justifying a delayed open release with safety considerations."
- Shares of the Hong Kong-listed company slid nearly 4pc at market close on Aug 14, despite the startup raising funds at nearly $7bn valuation earlier this year and operating a data center with at least 10,000 Chinese-made chips.
Insights by Ground AI
14 Articles
14 Articles
China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests
GLM-5.3 differs in that it is a general-purpose coding model that Z.ai says acquired cybersecurity capabilities through expanded post-training and reinforcement learning, rather than a purpose-built security system
·Chennai, India
Read Full Article
China’s AI startup Z.ai says open-source GLM-5.3 beats Anthropic’s Mythos 5 in cybersecurity test - Tech Startups
Less than two months after China’s Z.ai surprised the AI industry with GLM-5.2, a model that beat OpenAI’s GPT-5.5 on key benchmarks and narrowed the gap with leading U.S. AI labs, the Chinese startup is back with another challenge to […] The post China’s AI startup Z.ai says open-source GLM-5.3 beats Anthropic’s Mythos 5 in cybersecurity test first appeared on Tech Startups.
China's Z.ai says new model nears Anthropic's Mythos 5 in ...
·London, United Kingdom
Read Full Article
Coverage Details
Total News Sources14
Leaning Left1Leaning Right8Center2Last UpdatedBias Distribution73% Right
Bias Distribution
- 73% of the sources lean Right
73% Right
C 18%
R 73%
Factuality
To view factuality data please Upgrade to Premium
