U.S. media: Zhipu GLM-5.2 can match Mythos in vulnerability finding.

robot
Abstract generation in progress
Golden Finance reported on June 29, citing a comprehensive report by US tech media The Verge and The Wall Street Journal, that the domestic large model company Zhipu (02513.HK) recently released its open-weight model GLM-5.2. Although GLM still lags behind Anthropic and OpenAI's models on other more general tasks, some researchers claim that the model is comparable to Anthropic's Mythos model in certain vulnerability detection and cybersecurity scenarios.
According to data from cybersecurity company Semgrep, in some benchmark tests, GLM-5.2's performance surpassed the Claude Opus 4.8 model released by Anthropic in May.
Researchers pointed out that after receiving further instructions, Opus 4.8 and GLM-5.2 are comparable to Mythos in vulnerability discovery capabilities.
However, since GLM is an open-weight model, anyone can download and run it on off-the-shelf hardware. While this gives it great flexibility and allows advanced users deep access, it may also increase the risk of misuse by malicious actors. (Jinshi)
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned