← Back to live feed1 story
Five Fable 5.1 AI agents submitted five improved bounds for OpenAI’s integer multiplication result, using 92 sandbox workers and over 700 million tokens. The agents also reduced the Sidorenko residual constant from 978 to 189, reduced a unit distances proof grid to 12 edges, and found a new van der Waerden coefficient of 1/25. Separately, a mathematician alleged that OpenAI omitted credit in two manuscripts generalizing his earlier work.
Sign in to suggest edits
Key sources
- SOURCE@rohanarun“Five Fable 5.1 agents orchestrating a fleet of 92 sandbox workers spent over 700 million tokens with the following breakthroughs: 1) 5 new higher bounds for integer multiplication 2) Shrink the Unit distances grid to 12 edges 3) Cut the Sidorenko residual constant from 978 to 189 4) Found new Van der Waerden coefficient c = 1/25”x.com
- SUPPORT@0xdoug“Validated and merged Rohan's PR. Big gain on a really difficult regime (that frankly I was stuck at). Incredible work. κ = 2⁻¹⁵ (tightened from κ = 2⁻¹⁸²)”x.com
- SUPPORT@andrewcurran_“The latest redesigned the finite network to share intermediate computations and scratch space, then tightened the recursion and Gaussian estimates.”x.com
- SOURCE@himanshustwts“The result is also contained in a paper [8] released by OpenAI on October 6, 2026, in which the proof strategy and specific choices of notation are identical to a preliminary version of the present paper that was uploaded to ChatGPT on September 8, 2026.”x.com
- SUPPORT@intuitmachine“We have a new social issue here: any public proposal makes the proposer vulnerable to being scooped by AI.”x.com
- SOURCE@rohanpaul_ai“OpenAI withdraws some of the math paper that it published two days back.”x.com
- SUPPORT@pmddomingos“Surprise: the verified Navier-Stokes proof doesn’t match the one in the paper.”x.com
- SUPPORT@valeriocapraro“They identify at least two mismatches between the written intermediate results and their Lean counterparts: One estimate claims that four additional input derivatives suffice. The Lean version requires five: a weaker result. A pressure-flux estimate is obtained through a different bound, and proved through a different argument. It is not clear whether these mismatches invalidate the entire proof.”x.com