OpenAI posted 722 maths papers from one prompt to one agent.
OpenAI published 722 mathematics manuscripts on GitHub on Tuesday, all written by an internal model it has not released. A spokesperson said almost every result came from a single prompt handed to a single agent. Formal proof checks cover only 162 of the papers, and mathematicians call the one-prompt claim unverified.
Why it mattersThe cost of a frontier maths result just fell from a 10,000-agent swarm to three hours of compute. If that holds, every field with hard open problems is next, and checking the output stays unsolved.
Read in the app — free, in 9 languages
Related stories
The Erdős problems site stopped taking proofs after an AI flood.2026-10-07Twenty tries at the same task cut the best agent's score by a third.
2026-10-06A model learned to write without the method that trains every AI.
2026-10-06The US got 16 countries to put AI at the centre of science.
2026-10-05Mathematicians cracked five open problems using an ordinary chat box.
2026-10-05