OpenAI says an internal artificial intelligence model, described as more powerful than GPT-6 Astra, generated a proposed solution to one of mathematics’ seven Millennium Prize Problems. The company’s claim follows work involving 10,000 AI agents and a problem carrying a $1 million prize. However, mathematicians are questioning how independently the system reached the result. The debate centers on whether the proposed solution represents a genuinely original mathematical achievement or relies on existing knowledge and methods in ways that have not been fully established. CoinDesk reported the development on September 9, 2026. The report does not identify the specific Millennium Prize Problem or say that the proposed solution has been formally verified. As a result, the announcement remains a notable claim about the capabilities of large-scale AI systems rather than a confirmed mathematical breakthrough.
