OpenAI says it just logged hundreds more results on major math problems - and most of them came from a single prompt.
The company said it posted hundreds of new results this week, covering what it called major math problems. Per OpenAI, the majority of those results were generated in response to one prompt given to one AI agent, not a large team or a sprawling research push. The announcement did not name the specific problems, specify who checked the work, or explain how it defines 'major.' No other details were shared.
That's the real story here: a claim about leverage, not proof. One prompt producing hundreds of results would mean a single model query can now generate output at a pace that would once have taken a research group months. Whether that output holds up to scrutiny is a separate question entirely, and one OpenAI did not answer.
OpenAI has made confident claims about AI and math before, but this one comes with less verification detail than usual - treat 'hundreds of results' as a number to watch, not trust, until someone checks the math.