\n\n
OpenAI announced that its unreleased frontier model produced 372 new mathematical results, almost all derived from a single prompt to a single AI agent. This batch of findings, spanning long‑standing problems in mathematics and theoretical computer science, was generated far more efficiently than earlier large‑scale efforts such as the 10,000‑agent swarm that tackled the Navier–Stokes equations. While the company shared the volume of results, it has not released the specific prompts or the compute budgets per problem, despite calls from its advisory group for full transparency. The disclosure highlights rapid progress in AI‑driven theorem proving but also raises questions about reproducibility and openness in the emerging field of AI‑assisted mathematics. [[1, 2]]