Event ·Multi-source

GPT-6 Astra发布，多项基准测试创纪录

OpenAI发布GPT-6 Astra模型，在ARC-AGI 3测试中达到98.6%，远超5.6 Sol的7.8%。Exploit-Bench网络安全测试获得满分100%，FrontierMath Tier 4得分为97.6%，高于Fable 5.1的87.8%。DeepSWE v1.1测试得分为74.1%，优于Fable 5.1的67.4%。

Current read

OpenAI的GPT-6 Astra在数学、科学、网络安全等领域创下新纪录，Greg Brockman称其标志着AGI时代的到来。

11 sources93° AI heatLast updated 2026/9/3 18:07:36

Evidence

11 sources

Primary sourceThe Rundown AI

2026/9/3 18:07:36View source

Related source 1orange.ai

2026/9/3 23:36:56View source

Related source 2Sam Altman

2026/9/3 19:49:10View source

Related source 3Greg Brockman

2026/9/3 20:20:14View source

Related source 4OpenAI

2026/9/3 19:32:16View source

Related source 5elvis

2026/9/3 19:55:42View source

Related source 6Simon Willison’s Weblog

2026/9/3 20:18:41View source

Related source 7Gary Marcus

2026/9/4 01:34:52View source

Related source 8lmarena.ai

2026/9/5 17:40:41View source

Related source 9@OpenAIDevs

2026/9/3 19:33:24View source

Related source 10宝玉

2026/9/3 19:48:56View 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