PulseAugur
EN
LIVE 19:45:14
中文(ZH) 深信服AI在CyberGym评测跻身全球前四，国内第一

Sangfor AI ranks top 4 globally in AI security benchmark

By PulseAugur Editorial · [1 sources] ·

Sangfor AI has achieved a top-four global ranking and first place in China on the CyberGym benchmark for AI security capabilities. Utilizing the GLM-5.2 foundation model, Sangfor AI successfully addressed 1301 out of 1507 tasks, demonstrating an 86.3% success rate against a comprehensive set of ARVO and OSS-Fuzz vulnerability challenges. This performance highlights the growing strength of domestic AI models in specialized security evaluations. AI

IMPACT Demonstrates advancements in AI security capabilities, particularly with domestic foundation models.

RANK_REASON Performance on a specialized AI security benchmark. [lever_c_demoted from research: ic=1 ai=1.0]

Read on 36氪 (36Kr) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Sangfor AI ranks top 4 globally in AI security benchmark

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Performance on a specialized AI security benchmark. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, product
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
59 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

Full methodology in our editorial standards.

COVERAGE [1]

  1. 36氪 (36Kr) TIER_1 中文(ZH) ·

    Sangfor AI Ranks Top 4 Globally and First in China in CyberGym Evaluation

    36氪获悉，近日，在面向AI安全能力的国际基准测试CyberGym中，Sangfor AI在依托纯国产基座模型（GLM-5.2）的固定模型配置下，系统面对涵盖ARVO与OSS-Fuzz的全量高难度漏洞挑战，在1507项任务中成功解决1301项，整体成功率达到86.3%。最终整体评测成绩跻身全球前四，位列国内参评团队第一。

RELATED ENTITIES

RELATED TOPICS