OpenAI researcher on model solving Navier-Stokes: 'a different sport altogether', last 3 months 'hell'
Confident_Salt_8108 · reddit · 2026-09-29
An X post quotes an OpenAI researcher on frontier model performance on Navier-Stokes: "[it] surprised the fuck out of us" ... "Last 3 months = hell" ... "Suddenly we weren't dealing with just a small jump in capabilities; we were talking about a different sport altogether."
This hints at an unexpectedly large generational leap in math/physics reasoning, but the post names no specific model or benchmark; the source is a retweet by X user joedaroo and remains unverified.
Related event: OpenAI's swarm of agents tackles Navier-Stokes, stirring math debate(2 posts)→
More from Models
- Dev: The Era of Subsidized AI Tokens Has Begun Its Descent — RileyRalmuto · 2026-09-29
- Voice model Samuel learns human-like mouth positioning without explicit training — deliprao · 2026-09-29
- OpenAI brings back $200 Pro plan, but Astra effectively gets its own tighter usage limits — cedric_chee · 2026-09-29
- Claude goes down as users joke the AI refuses 'secretary work' — neil_chilson · 2026-09-29
- OpenAI scraps GPT-6.1 release after safety testing shows regression, deception and unsafe tool use — Ars Technica AI · 2026-09-29
- Gemini 4 Pro spotted on LMArena under codename 'Gemini 3.8 Flash', overthinks — XCyberThunder · 2026-09-29