OpenAI's GPT-6 Astra on ARC-AGI-3
What happened
OpenAI unveiled GPT‑6, codenamed Astra, and tested it on the ARC‑AGI‑3 benchmark, a challenging suite designed to measure reasoning and problem‑solving abilities that approach human‑level intelligence. Astra achieved a new state‑of‑the‑art score, outperforming GPT‑4 and other leading models across most tasks, demonstrating markedly better abstract reasoning, few‑shot learning, and transfer capabilities. The release showcases a significant leap in the scalability and versatility of large language models toward more general AI behavior.
Why it matters
The breakthrough suggests the AI field is closing the gap to artificial general intelligence, prompting faster competition among labs and raising urgent questions about safety, alignment, and responsible deployment.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from OpenAI
Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026OpenAI launches ChatGPT Wallet, AI agents pay for purchaseOpenAI · 13 Nov 2026OpenAI disbands its robotics team as its general models start running robotsOpenAI · 13 Nov 2026Yang-Mills, OpenAI proof, and the mass gap problemOpenAI · 13 Nov 2026This page was collected from arcprize.org and summarised automatically by AIPulsee's pipeline. See how it runs.