OpenAI repo claims GPT-6-Astra Lean-formalized a prime-gaps bound of 186
scaling01 · x · 2026-09-04
A new OpenAI GitHub repo, PrimeGaps186, claims a Lean 4 formalization by GPT-6-Astra proving infinitely many consecutive prime pairs with gaps of at most 186. Key points:
- The repo includes Lean code plus a Python numerical certificate, deriving DHL[40,2]: every admissible set of 40 shifts has infinitely many translates containing at least two primes
- The result is conditional on three explicit input axioms — the cited estimates and numerics are not yet Lean-proved
- The name echoes the gpt-6-astra traces found in Codex, further corroborating the codename rumor
It reads more as a demo of frontier-model math formalization than a new theorem.
Related event: GPT-6-Astra Proves Prime Gaps ≤186 in Lean as Records Fall in Days(22 posts)→
More from Models
- GPT-6 Astra and Claude Fable 5.1 write $10,000 research papers — experts reject all — sayashk · 2026-10-21
- GitHub Copilot's HydraFusion Orchestrates Multiple Models Per Task — WirelessLife · 2026-10-21
- OpenVLHarness turns any MLLM into a visual reasoner, +9.9 points across 22 datasets — yining_hong · 2026-10-21
- User claims Gemini analyzed his tennis match video and improved his game — zacharynado · 2026-10-21
- Opus 5.5 tops OpenRouter spending in 20 of 29 task categories — mmurph · 2026-10-21
- Meta team builds open-source 30B LLM from scratch in months, hails 'second era' — TimDarcet · 2026-10-21