ARTFEED — Contemporary Art Intelligence

Chinese AI model GLM-5.2 matches US rival in cybersecurity tests

ai-technology · 2026-06-29

A Chinese AI model, GLM-5.2 developed by Beijing-based Zhipu AI, has demonstrated competitive performance against US rival Anthropic's Mythos model in cybersecurity tasks, marking what researchers call another 'DeepSeek moment'. Released on June 13, GLM-5.2 outperformed Anthropic's Claude Opus 4.8 in benchmarking tests conducted by cybersecurity company Semgrep, as reported by The Wall Street Journal. After receiving further instructions, GLM-5.2 matched the US model's capabilities, narrowing the gap between Chinese and American AI in bug-hunting prowess.

Key facts

  • Zhipu AI's GLM-5.2 was released on June 13.
  • GLM-5.2 beat Anthropic's Claude Opus 4.8 in Semgrep tests.
  • The achievement is hailed as another 'DeepSeek moment'.
  • GLM-5.2 matched Anthropic's Mythos model after further instructions.
  • Semgrep is a cybersecurity company that conducted the benchmarking.
  • The Wall Street Journal reported the results.
  • Zhipu AI is based in Beijing.
  • The competition narrows the AI gap between China and the US.

Entities

Institutions

  • Zhipu AI
  • Anthropic
  • Semgrep
  • The Wall Street Journal

Locations

  • Beijing
  • China
  • United States

Sources