Model Releases

Greg Brockman shares post: GPT-6 Astra reaches SOTA on ARC-AGI-3 as benchmark nears saturation

Original title:arc-agi-3 is now saturated

AI summary

Greg Brockman shared an evaluation by @arcprize saying OpenAI's GPT-6 Astra achieved SOTA on ARC-AGI-3, which he described as saturated. Astra scored 63% with the standard harness and 99% with a new Provider Adapter harness, surpassing human performance on 96% of ARC-AGI-3 levels; the leaderboard chart also shows higher reasoning levels are often cheaper because Astra completes levels in fewer actions, reducing model calls and tokens.

Key Takeaways

  • Greg Brockman shared an evaluation by @arcprize saying OpenAI's GPT-6 Astra achieved SOTA on…
  • Astra scored 63% with the standard harness and 99% with a new Provider Adapter harness,surpassing…
  • the leaderboard chart also shows higher reasoning levels are often cheaper because Astra completes…

Quick Facts

Published
Updated
Source
X:Greg Brockman (@gdb)
Category
Model Releases
AI score
72
Read the original Read the full article at the original source
Why it matters

转发 ARC Prize 对 GPT-6 Astra 的评测数据,标准与 Provider Adapter 两种 harness 分差大,可据此了解 harness 对得分的影响。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Greg Brockman shares post: GPT-6 Astra reaches SOTA on ARC-AGI-3 as benchmark nears saturation. September 4, 2026. https://new.nat6.net/en/news/cmtm2suk00171rooej14mutgj/

View all →