OpenAI's Astra Model Surpasses Humans in Spatial Reasoning

OpenAI's head of applied research, Boris Power, recently stated on X that spatial reasoning was one of the few areas where humans significantly outperformed computers, but the Astra model has closed that gap. Independent AI researcher @spicylemonade reported that Astra performed flawlessly in his spatial reasoning tests, getting all questions correct, whereas all previous language models made mistakes. Astra is OpenAI's latest model, released about a year after the GPT-5 series and about two months after GPT-5.6. OpenAI calls it the "world's most intelligent and aligned model," capable of filling out forms, formatting documents, scheduling appointments, and operating software. OpenAI president Brockman said anything you can do on a computer, Astra can do quickly, adding "Welcome to the AGI era."