Event ·Multi-source

新模型在 CVE 漏洞复现基准上追平 DeepSeek V4 Pro 0813

在 32 个基准 CVE 的复现测试中，该模型以 pass@3 找回 28 个，比例 87.5%，与 DeepSeek V4 Pro 0813 并列第二。榜首是 GPT-6 Astra，成绩为 90.6%。发帖人指出 DeepSeek V4 Pro 0813 的成绩不可能是从 Astra 蒸馏得来，认为其优势来自数据规模。

Current read

一个模型在漏洞复现测试里找回 87.5% 的 CVE，跟 DeepSeek V4 Pro 0813 打平，只差 GPT-6 Astra 三个百分点，安全研究方向可以看看。

10 sources42° AI heatLast updated 2026/10/22 14:00:18

Evidence

10 sources

Primary sourceTeortaxes

2026/10/22 14:00:18View source

Related source 1Artificial Analysis

2026/10/21 15:53:56View source

Related source 2Decoder

2026/10/20 16:40:29View source

Related source 3IT之家

2026/10/20 23:45:48View source

Related source 4pandaily

2026/10/21 01:55:43View source

Related source 5arXiv: DeepSeek

2026/10/21 04:09:37View source

Related source 6rohanpaul_ai

2026/10/21 07:56:47View source

Related source 7theverge

2026/10/21 13:57:43View source

Related source 8PolymarketMoney

2026/10/21 15:49:50View source

Related source 9TestingCatalog

2026/10/21 18:57:16View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story