OpenAI just nailed ten tough math and computer science problems that had stumped researchers for years. Using a secret internal AI model called Astra, the company managed to find solutions that had resisted human minds for a decade or more. The whole effort cost roughly $2,000 in token fees a bargain for breakthroughs of this scale.
But OpenAI isn’t stopping at results. They’ve also opened up ChatGPT’s most advanced models to 100,000 academic researchers at no charge. This move aims to democratize access to cutting-edge AI tools, empowering scientists and mathematicians to tackle unsolved problems without pricey subscriptions standing in the way.
Astral Leap in AI-Assisted Mathematics
Astra’s success includes disproving the Erdős unit-distance conjecture, a long-standing puzzle in geometry. From there, it tackled open questions spanning eight fields of mathematics and theoretical computer science. Human experts transformed Astra’s AI-generated proofs into formal manuscripts, which were then verified using Lean, a proof-checking software.
OpenAI’s initiative signals a shift in how AI interacts with research: from novelty to fundamental infrastructure. The company bets that free access will quicken progress, as more researchers get hands-on with these smarter tools. This approach could reshape academic workflows by making powerful AI-driven mathematical reasoning readily available, sidestepping costly barriers that stifle innovation.
This article is informational and doesn’t constitute financial advice.
