As of: 2026-10-15 10:09 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/v/zeta-explained-openai-zero-free-region-7-8/

# Zeta Explained #116.5: OpenAI's Zero-Free Region Re(s)>7/8

Zeta Explained, 7 October 2026, YouTube. 47,911 views as of 15 October 2026.
Kind: Tutorial.
Watch: https://www.youtube.com/watch?v=XtHsE99VGpk

## Why it is here

A math-explainer series (9:25) on how OpenAI's claimed zero-free region Re(s) > 7/8 differs from known zero-free regions and what it implies for the error term of the prime number theorem. 48k views.

## Description (written by Gemini from the video)

**Summary**  
This video is an educational breakdown by the creator of the YouTube channel *Zeta Explained* discussing OpenAI’s claimed mathematical paper on the quasi-Riemann hypothesis. The presenter reviews traditional zero-free regions for the Riemann zeta function, explains what OpenAI’s announced zero-free half-plane $\text{Re}(s) > 7/8$ represents, and outlines its theoretical consequences for bounding the error term in the Prime Number Theorem.

**What is shown**  
- **[00:00]** Title slide introducing episode 116.5 on OpenAI’s new zero-free region $\text{Re}(s) > 7/8$.
- **[00:14]** The title and abstract page of OpenAI’s paper titled *"The Quasi-Riemann Hypothesis: A Zero-Free Half-Plane $\text{Re}\,s > 7/8$"*, dated 30 September 2026.
- **[00:42]** Complex plane plots displaying non-trivial zeros along the critical line $\text{Re}(s) = 1/2$ within the critical strip $0 < \text{Re}(s) < 1$, alongside the classical zero-free boundary.
- **[02:27]** Comparative plots contrasting historical zero-free regions—de la Vallée Poussin (1899) and Vinogradov–Korobov (1958)—with OpenAI’s announced constant boundary at $\sigma > 7/8$ (confining zeros to $1/8 \le \sigma \le 7/8$ by functional symmetry).
- **[04:45]** Graphs of the prime-counting step function $\pi(x)$ compared against the logarithmic integral $\text{Li}(x)$.
- **[05:49]** Equivalence theorem between the zero-free region $\text{Re}(s) > B$ and prime error bounds $\pi(x) = \text{Li}(x) + O(x^B \log x)$, displaying error bands comparing Vinogradov–Korobov’s error bound to the tighter $O(x^{7/8} \log x)$ bound implied by OpenAI's claim.
- **[08:38]** Slide asking *"How does the proof work?"*, where the presenter explains that the paper is nearly 200 pages and still being analyzed by the mathematical community.

**Claims & numbers**  
- The presenter notes that OpenAI’s paper is dated September 30, 2026, though the material was made public on October 6, 2026 [00:18].
- The abstract shown claims a proof that all finite-order Hecke $L$-functions over $\mathbb{Q}(\sqrt{-3})$ and all Dirichlet $L$-functions are zero-free for $\text{Re}(s) > 7/8$ (allowing the principal pole at $s = 1$), thereby proving the quasi-Riemann hypothesis [00:14].
- By functional symmetry, the presenter states that this result confines all non-trivial zeros of $\zeta(s)$ to the strip $1/8 \le \sigma \le 7/8$ [04:08].
- The presenter notes the result implies an error bound for the Prime Number Theorem of $E(x) = O(x^{7/8} \log x)$, improving upon the Vinogradov–Korobov error bound [07:13].
- The presenter states that OpenAI’s proof spans nearly 200 pages [04:32, 09:00].

**Notable quotes**  
- *"The main thing is that the $L$-functions are kind of extensions or generalizations of the Riemann zeta function... the main thing to know... is that it says the Riemann zeta function is zero-free in this half-plane."* [00:25]
- *"So if you believe the result... there can not be any zeros with a real part greater than 7/8ths."* [03:28]
- *"How does the proof work? ... the answer is: I have no idea, and I would guess most people have no idea... the proof is nearly 200 pages."* [08:39]

**Assessment**  
This is an independent educational commentary and mathematical explainer video rather than an official product launch or demonstration. The presenter does not verify or independently check OpenAI's ~200-page proof, explicitly noting that it remains unverified by the presenter and is still being digested by the broader mathematical community.

_Described by gemini-3.8-flash on 2026-10-15 from the video's audio and frames._

## Related

- 2026-10-06: [OpenAI model claims the quasi-Riemann hypothesis](https://postcutoff.com/e/2026-10-06-quasi-riemann-hypothesis-openai/)