Event ·Multi-source

OpenAI的Astra模型在ARC-AGI-3基准上取得突破

OpenAI的GPT-6 Astra模型在ARC-AGI-3基准上得分达到63%。通过新的provider adapter harness，该模型得分提升至99%。Astra在ARC-AGI-3的96%关卡上超越了人类表现。该模型构建了我们见过的最精确的新环境符号模型。

Current read

OpenAI的Astra在ARC-AGI-3上达到99%分，96%关卡超越人类，表现惊艳。

11 sources73° AI heatLast updated 2026/9/3 21:46:00

Evidence

11 sources

Primary sourceGreg Brockman

2026/9/3 21:46:00View source

Related source 1Gary Marcus

2026/9/3 21:34:16View source

Related source 2The Rundown AI

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

Related source 3OpenAI

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

Related source 4Sam Altman

2026/9/3 19:49:10View 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 7orange.ai

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

Related source 8歸藏(guizang.ai)

2026/9/4 00:59:38View source

Related source 9Latent Space (swyx)

2026/9/4 05:18:11View source

Related source 10向阳乔木

2026/9/4 08:57:07View 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