HEXAGONbeta

First Proof Third Batch

Primarily human-written textHuman understanding: all partscs.AI — Artificial Intelligencemath.GM — General Mathematics

Contributed by Mohammed Abouzaid, Nikhil Srivastava, Rachel Ward, Lauren Williams ↗

Version 1 / Oct 14, 2026 / CC BY 4.0

Abstract

To assess the ability of current AI systems to correctly solve research-level mathematics problems, we tested several AI systems on our ``third batch'' of ten problems in a broad range of mathematical fields; these problems arose naturally in the research process of the contributors. This document includes the problems, our methodology, and the results of our testing. We assessed the AI-generated solutions based on mathematical correctness, quality of exposition, and quality of attribution and citations. We provide links to supplementary documents including the human solutions, the AI-generated solutions, and the referee reports and logs for the AI-generated solutions.

References

  1. Abouzaid, Mohammed; Blumberg, Andrew J.; Hairer, Martin; Kileel, Joe; Kolda, Tamara G.; Nelson, Paul D.; Spielman, Daniel; Srivastava, Nikhil; Ward, Rachel; Weinberger, Shmuel; Williams, Lauren. First Proof. 2026. arXiv preprint \texttthttps://arxiv.org/abs/2602.05192arXiv
  2. Abouzaid, Mohammed; Srivastava, Nikhil; Ward, Rachel; Williams, Lauren. First Proof Second Batch. 2026. arXiv preprint \texttthttps://arxiv.org/abs/2606.18119arXiv
  3. Alexeev, Valery; Brunyate, Adrian; Engel, Philip. Compactifications of moduli of elliptic K3 surfaces: Stable pair and toroidal. Geometry & Topology, vol. 26, no. 8, pp. 3525–3588. 2022
  4. Bischoff, Manon; Howlett, Joseph. AI solves 79-year-old mathematical mystery. 2026. Scientific American, https://www.scientificamerican.com/article/ai-solves-79-year-old-math-mystery-of-six-dimensional-spheres/link
  5. Cohen, Albert; DeVore, Ronald; Schwab, Christoph. Analytic regularity and polynomial approximation of parametric and stochastic elliptic PDE's. Analysis and Applications, vol. 9, no. 1, pp. 11–47. 2011
  6. Giovenzana, Luca; Thompson, Alan. Degenerations and fibrations of K3 surfaces: Lattice polarisations and mirror symmetry. 2024arXiv
  7. Kakaes, Konstantin. AI Has Solved One of Math's $1 Million Millennium Prize Problems. 2026. Quanta Magazine, https://www.quantamagazine.org/ai-has-solved-one-of-maths-1-million-millennium-prize-problems-20260908/link
  8. Kimi Team, Moonshot AI. Kimi K3 Technical Report. 2026link
  9. Lee, Melissa. `hello there the jacobian conjecture is false thanx': why a tiny social media post has mathematicians rethinking AI. 2026. The Conversation, https://theconversation.com/hello-there-the-jacobian-conjecture-is-false-thanx-why-a-tiny-social-media-post-has-mathematicians-rethinking-ai-283883link
  10. Moonshot AI. Kimi K3 model weights. 2026. Revision f831ab66814297da540d832a5235f8e904f29d06. Accessed October 2026. Hugging Face, https://huggingface.co/moonshotai/Kimi-K3/tree/f831ab66814297da540d832a5235f8e904f29d06link
Show all 15 referencesShow fewer references
  1. OpenAI posts 700 maths preprints online: mathematicians are up in arms. 2026. Nature (news), https://www.nature.com/articles/d41586-026-03196-8link
  2. Nickl, Richard. Bayesian Non-linear Statistical Inverse Problems. EMS Press. 2023
  3. RadixArk. Kimi-K3-DSpark draft model. 2026. Accessed October 2026. Hugging Face, https://huggingface.co/RadixArk/Kimi-K3-DSparklink
  4. Stuart, Andrew M. Inverse problems: a Bayesian perspective. Acta Numerica, vol. 19, pp. 451–559. 2010
  5. William P. Thurston. On proof and progress in mathematics. Bulletin of the American Mathematical Society (N.S.) , 30(2):161--177, 1994.

Version history

  1. v1Submitted by Lauren WilliamsInitial depositCurrentOct 14, 2026