The Artificial Intelligence Show · Tuesday, September 8, 2026
Recent advancements in AI models, exemplified by GPT-6 Astra, demonstrate a significant leap in problem-solving capabilities. Astra achieved a 99.9% score on the ArKI-3 benchmark, which tests an AI's ability to learn and strategize in unknown scenarios without explicit rules. Furthermore, it scored 97.6% on a challenging mathematics test, indicating improved reasoning for scientific discovery.
“Notably, learning how to solve unfamiliar problems.”
“Astra scored a 99.9% on that under OpenAI's evaluation setup.”
“In mathematics, Astra scored 97.6% on Frontier Math's hardest tier. That's up from 83%.”