The 867 KB PDF uploaded to OpenAI's math GitHub repository on October 25 2026.
OPENAI PREPRINT CLAIMS EXPONENTIAL TIME HYPOTHESIS FALSE
*A new 867 KB PDF posted on OpenAI’s math GitHub asserts the long‑standing Exponential Time Hypothesis is false. The claim, dated October 25 2026, could upend decades of complexity theory research.*
OpenAI’s math team has dropped a bombshell on the theory community. A 867 KB PDF titled “The Exponential Time Hypothesis Is False” appeared on the organization’s GitHub repository on October 25 2026. The paper declares the long‑standing Exponential Time Hypothesis, a cornerstone of computational complexity, to be false. No peer‑review badge, no journal citation—just a raw preprint available to anyone with a browser. The timing and venue amplify the shock: a leading AI lab, a public code host, and a claim that could rewrite textbooks. The stakes are high; ETH underlies countless hardness proofs and cryptographic assumptions. If the argument holds, the ripple effect will touch algorithms, security, and the economics of computing.
What does the new preprint actually say?
The document, titled “The Exponential Time Hypothesis Is False,” is a 867 KB PDF uploaded to the openai/math repository on GitHub. Its sole claim is that the Exponential Time Hypothesis (ETH) does not hold. No additional context, experimental data, or external citations appear in the file listing. The paper’s presence on a public code platform makes it instantly accessible to researchers worldwide. By posting the manuscript as a raw PDF, OpenAI bypasses traditional peer‑review channels, forcing the community to evaluate the argument on its own merits.
Why is the ETH important to computer science?
ETH underpins many hardness results in computational complexity. It posits that solving 3‑SAT requires exponential time in the worst case. If the hypothesis fails, a cascade of conditional lower bounds could collapse, reshaping expectations for algorithmic performance. The hypothesis also guides resource allocation in cryptography, optimization, and theoretical research. A credible refutation would force textbooks to rewrite core sections and could accelerate efforts to find faster algorithms for problems previously deemed intractable.
How are experts reacting to the claim?
The preprint’s release on a public repository has already sparked discussion on forums such as Hacker News. Commentators note the rarity of a claim that overturns a foundational hypothesis without accompanying peer review. Some analysts caution that the community will demand a rigorous proof, replication attempts, and verification of every logical step. Others see the move as a strategic signal from OpenAI, highlighting its ambition to influence frontier theory. The debate is unfolding in real time, with no consensus yet reached.
What could happen if the claim survives scrutiny?
Should the proof withstand expert examination, the immediate effect would be a reassessment of many conditional results that rely on ETH. Researchers might revisit algorithmic lower bounds, potentially unlocking new approaches to NP‑hard problems. Funding bodies could shift priorities toward exploring the newly opened landscape. Conversely, a refutation of the claim would reinforce ETH’s standing and reaffirm the current theoretical framework. Either outcome will reverberate through academic conferences, grant proposals, and industry roadmaps that depend on complexity assumptions.
Frequently asked questions
What is the title of the new OpenAI preprint?
The preprint is titled “The Exponential Time Hypothesis Is False.”
When was the preprint uploaded to GitHub?
The PDF was uploaded on October 25 2026.
Where can the preprint be accessed?
It is available on the openai/math GitHub repository at the provided URL.
The community now faces a choice: scrutinize the proof line by line, or dismiss it as premature hype. Either path will generate new research, new debates, and new data points for the field. OpenAI has forced the issue onto the open road, and the next weeks will determine whether the claim reshapes complexity theory or fades into the archive of bold but unverified conjectures.
Sources: [Hacker News post linking to https://github.com/openai/math/blob/main/preprints/The-Exponential-Time-Hypothesis-is-False-October-25-2026/paper.pdf]