GPT-5.6 Sol reportedly disproves a 30-year-old statistics conjecture in 90 minutes after humans couldn't crack it

A recent report highlights a significant achievement by OpenAI’s GPT-5.6 Sol Pro, which successfully disproved a longstanding conjecture related to the Benjamini-Hochberg method in statistics. This breakthrough was accomplished in approximately 90 minutes by a statistics professor at the University of Pennsylvania, contrasting sharply with the performance of its predecessor, GPT-5.5, which failed to find a solution even after 20 hours of computation. This rapid advancement raises intriguing questions about the capabilities of AI in generating new knowledge versus merely recombining existing information.

The conjecture in question has been a central topic in statistical theory for over three decades, and the ability of GPT-5.6 Sol Pro to resolve it so quickly suggests a potential leap in the application of AI in complex problem-solving scenarios. The findings prompt further investigation into whether AI can indeed create novel insights or if it is limited to synthesizing previously acquired knowledge. This development not only showcases the evolving capabilities of AI models but also emphasizes the ongoing discourse regarding the nature of intelligence and creativity in artificial systems. For more details, refer to the original article on The Decoder.