今天新增gemini-4IMAGE
Chrome 团队称 Gemini 4 Carbon 安全补丁生成超对手 10 倍
Google 称 Carbon 生成高质量安全补丁的能力超另一实验室前沿模型 10 倍,安全代码成功率超 Argon 两倍,提示注入测试零攻破,AA 得分 59 比 58 压过 Opus 5.5。
Choblin
@choblin294 小时前收录原推发布于 2026.10.29

Choblin @choblin29原推发布于 2026.10.29
Holy shit. Google's Chrome team says Gemini 4 Carbon performed over 10X better than another lab's large frontier model at generating high-quality security patches. Google also reports over 2X higher secure-code success than Argon, nearly 80% fewer high-severity misalignment alerts, and ZERO successful attacks on its evaluated prompt-injection benchmark. And Carbon is beating Claude Opus 5.5 on Artificial Analysis, 59 to 58. I genuinely cannot wait to try this model. I'm VERY, VERY excited.
这条是引用着另一条写的:「Logan Kilpatrick:Gemini 4 Carbon 在编程与对齐上大幅超越 Argon」
x.com/choblin29/status/2115494150690984377采集于 2026.10.29 10:01