On November 15, 2026, OpenAI published a guest post by physicist Nathaniel Craig describing how he worked with GPT-6 Astra on a candidate way to define the Standard Model of particle physics on a lattice, a discrete grid that stands in for spacetime. The post is the first on a new OpenAI science blog launched the same day. Craig, who leads the particle theory group at the University of California, Santa Barbara and is a permanent member of its Kavli Institute for Theoretical Physics, is a visiting scientist at OpenAI. The account is based on his October 6, 2026 preprint, "A gauge-invariant measure for lattice chiral fermions" (arXiv:2610.09093), which lists affiliations at UC Santa Barbara, the Kavli Institute, and OpenAI.
The Standard Model describes the known elementary particles and three of the four fundamental forces, and it has been tested across an enormous range of energies. Craig's point is that physicists still lack a nonperturbative definition of the full theory, meaning a definition that does not treat the interactions as small corrections. Lattice field theory has been highly successful for the strong force, including first-principles calculations of the proton's mass. Extending that approach to the whole Standard Model is much harder because the weak force treats left-handed and right-handed fermions differently. A straightforward lattice version tends to produce extra "doubler" particles, so the particle content of the weak force comes out wrong.
Craig writes that earlier work, including overlap fermions developed by Rajamani Narayanan and Herbert Neuberger and a phase prescription by Martin Lüscher for simpler theories such as a chiral version of electromagnetism, left a gap for the full Standard Model that had stood for more than 25 years. His strategy was to represent the same force fields on a succession of finer grids and to build the needed phase rule by comparing the allowed fermion states on neighboring grids. He says he had not been able to prove that the idea worked in all the cases it had to cover. According to the post, GPT-6 Astra helped turn that strategy into a construction, suggested how to carry the force fields onto finer grids, showed that the comparisons would converge, and took the lead in developing the detailed proof.
The preprint's abstract states that the work constructs a gauge-invariant measure for overlap fermions on four-dimensional Euclidean lattices in a class of nonabelian chiral gauge theories that includes the Standard Model, and that this "provides a candidate nonperturbative lattice definition of the Standard Model." Craig says lattice field theorists Maarten Golterman, Andreas Kronfeld, and Yigal Shamir reviewed Astra's proof and asked questions that helped him gain confidence in the result. The paper also says Lean verification of the main proof is reserved for supplementary material.
Craig is careful not to treat the construction as a finished solution. He calls it a beginning rather than an end, "far more an answer in principle than an answer in practice," and says much work remains before it would be useful for realistic lattice simulations. He flags an open question about whether the construction satisfies reflection positivity at finite lattice spacing, a condition that helps connect lattice calculations to a quantum theory in which probabilities are conserved. He also notes that defining the interacting theory in continuous spacetime remains open, pointing to OpenAI's recent work on the existence of Yang-Mills theory as a related piece rather than as a substitute for expert review of this paper.
The November 15 post is a first-party account of an October preprint, not a peer-reviewed journal article, and the scientific claims should be read with that in mind. OpenAI also posted a companion note the same day introducing the science blog itself. That post is a publishing venue, not a separate research result.