OpenAI推出网络安全专用模型GPT-6 Cyber
Start here
2026年11月6日，OpenAI正式推出网络安全领域专用智能体模型GPT-6 Cyber，旨在推进具备安全对齐特性的网络安全智能体能力。该模型在提升漏洞挖掘能力的同时，将自主越界攻击其他平台的概率降至1%以下。基准测试方面，GPT-6 Cyber在ExploitGym漏洞利用测试中取得48.5%的得分，较此前GPT-5.6 Cyber的32.1%显著提升；此外，该模型已协助发现了Firefox浏览器的81个未知漏洞。
AI-generated from reports · Updated 2 hours ago
Report timeline
Follow the reports to see different sides of the story.
- Hacker News · AISelectedOpenAI Unveils GPT-6 Cyber to Advance Cybersecurity Agents
OpenAI has introduced GPT-6 Cyber, an agentic model focused on cybersecurity. The model aims to advance the capabilities of safety-aligned cybersecurity agents.
- The Verge · AI 筛选SelectedOpenAI launches GPT-6 Cyber, a dedicated cybersecurity model
OpenAI has officially launched GPT-6 Cyber, a cybersecurity AI model that enhances vulnerability discovery while reducing the probability of autonomous, out-of-bounds attacks on other platforms to below 1%. The model scored 48.5% on the ExploitGym benchmark—up from 32.1% achieved by GPT-5.6 Cyber—and helped uncover 81 previously unknown vulnerabilities in Firefox.
Story heat trend
Not enough continuous data yet to draw a trend.