OpenAI says its internal models produced results on all-dimensional local smoothing, the Exponential Time Hypothesis, and quantum-field-theory existence. Two writeups and Lean files are public; the physics work remains under academic review and is not yet released.
Current score
+3.98
30 base · Historic (tier 4 of 5, 30 pts)
× 0.9000 attribution · Primary causal contribution
× 0.3500 evidence · Technical disclosure or preprint
× 0.5500 realization · Complete public manuscript
× 0.9000 durability · Effectively permanent
Event-level product before credit split: 4.68
Historic potential (tier 4, base 30) for a release claiming advances on three major open questions across analysis, computational complexity, and quantum field theory. The score treats this as a provisional claim cluster, not as three accepted theorems or a completed Clay Millennium problem. Attribution is high because OpenAI attributes the results to its internal models (0.90); the models receive an 85% editorial credit share, with human prompting, existing mathematical foundations, and expert-led interpretation recognized. Evidence is 0.35 because the announcement is first-party and two results have public writeups and Lean formalizations, but the work is not peer reviewed or independently accepted and the physics artifacts are withheld. Realization is 0.55 because two formalized research outputs are public while the third remains unavailable for inspection. Durability is 0.90 conditional on correctness. Jev 1.14.0 classified the story as AI-relevant and novel, but assigned low probability to a notable outcome (0.13) and selected neither ledger (0.99), reflecting the ingest classifier's strict realized-outcome gate. Editorial inclusion here records the concrete publication of two public research outputs and an announced third claim, with explicit provisional status. Jev did not choose the rubric factors. Score: 30 × 0.90 × 0.35 × 0.55 × 0.90 × 0.85 = 3.9761 points.
What happened
On 26 October 2026, OpenAI announced three research results it attributes to its internal models. In harmonic analysis, it claims a proof of the local smoothing conjecture in every dimension, with implications for Kakeya and Stein restriction conjectures. It says the natural-language writeup and a Lean formalization are public. In theoretical computer science, it claims a refutation of the Exponential Time Hypothesis (ETH), a conjecture used to establish conditional hardness results. OpenAI published a writeup and Lean formalization and says a group led by ETH's original proposers Russell Impagliazzo and Ramamohan Paturi, with Ryan Williams, will organize an open examination. The organizers say their involvement is not certification.
For physics, OpenAI claims a mathematically rigorous existence proof for certain four-dimensional quantum field theories, including QCD and a completion of the Standard Model. It says the work includes a proof of pure Yang-Mills existence, which is part of the Clay Millennium Prize problem, but does not claim to have proved the full Yang-Mills mass-gap problem. The collaborating physicists have asked to hold the AI-produced artifacts while they prepare their own account.
This record covers the 26 October announcement as one cross-disciplinary research release, not three independently validated theorems. The local-smoothing and ETH manuscripts are public and include Lean formalizations, but neither is peer reviewed or independently accepted. The physics claim cannot yet be inspected through public source artifacts. OpenAI says it does not seek academic credit for the discoveries and that expert collaborators will work independently. The Reddit post is discussion and amplification, not evidence for the mathematical claims.
