WHAT NEW IN X10.29 THU

X 上的 AI 合集档案 · 22 个合集 · 60,636 条

今天新增gemini-4POST

Gemini 4 Carbon 距 Argon 仅两周:安全基准与开放节奏

Carbon 在 SEC-bench Pro 拿 88.2%、ExploitGym 54.8%、CWE-bench 71%,评测无护栏,定价与 Argon 同为 $2/$10,本周先向部分网络安全防御者开放。

Hybrid 🏃🏾
@markfenner6 小时前收录原推发布于 2026.10.29
Hybrid 🏃🏾 @markfenner原推发布于 2026.10.29
在 X 上打开原推

Google’s not playing around. Gemini 4 Carbon comes just two weeks after Argon. Google says its teams have had it working through engineering tasks for DAYS, tracking down and patching subtle bugs across large codebases. The cybersecurity numbers they’re reporting: • 88.2% on SEC-bench Pro for finding vulnerabilities • 54.8% on ExploitGym for validating exploits • 71% on CWE-bench for patching vulnerabilities Those evaluations ran without guardrails. Same introductory pricing as Argon: $2 input / $10 output per million tokens. Access starts with selected cyber defenders this week, with broader availability coming later.

这条是引用着另一条写的:「Google DeepMind 官宣 Gemini 4 Carbon 为下一代前沿模型」

x.com/markfenner/status/2115491821618905467采集于 2026.10.29 10:08