VibeMathedMath problems solved with AI

Bounded prime gaps: H1236H_1 \le 236

Write H1=lim infn(pn+1pn)H_1 = \liminf_{n\to\infty}(p_{n+1}-p_n). Announced on 1 September 2026: H1236H_1 \le 236, building on Stadlmann's 240240 of the previous day.

Result
Proved(see note)
Status
Candidate (review pending)
AI contribution
AI-assisted
Method
Argument
Field
Analytic number theory
Posed by
Alphonse de Polignac (the twin prime conjecture); the bounded form since Goldston, Pintz and Yıldırım
Year posed
1849
Years open
177y
Solved
2026-09-01
Model
unspecified AI agents
Vendor
Collaborators
Shiva Kintali
Verification
Unreviewed
Publication
Announced
Significance
62 / 100
Disclosed cost
Wikipedia
No dedicated article

What was actually shown

H1236H_1 \le 236, between Stadlmann's 240240 of 31 August and Axiom Math's 212212 of 3 September. Superseded within two days, which Kintali noted himself.

What the AI did

Kintali's own words on X: "I showed that H1 <= 236 with the help of AI, building on Julia Stadlmann's work." He describes the method as the GPY approach with the Maynard-Tao sieve, and says he then stopped his AI agents and moved to another problem. That is the whole of the public disclosure: no model is named, the division of labour is not described, and there is no preprint.

Verification

The thinnest source of the three 2026 bounded-gaps entries, and recorded as Candidate for that reason. An announcement on X with no preprint, no named model and no argument in public. It is recorded because a bound is a bound and this one held the record for two days between Stadlmann's 240 and Axiom Math's 212; it is not recorded as settled. If a preprint appears, this entry should be revisited.

Sources

Changelog1 change

Discussion