🥺 Sam Altman says AI reasoning is moving into a new league

In a conversation with Salesforce CEO Marc Benioff, Sam Altman described a striking progression in AI’s mathematical abilities.

GPT-5.5 was compared to an average math professor, GPT-5.6 to the top 1–2%, and Astra to a level slightly beyond that. Altman reportedly said OpenAI’s next internal model was tackling problems that even top mathematicians struggle with.

If these comparisons hold up, the bigger story isn’t any single benchmark. It’s how quickly the bar keeps rising.

AI is moving beyond answering questions toward solving problems that have challenged experts. The next generation could be a very different beast.

@aitopics