1,300 H200s + automated labs: Fedus's Neon model beats GPT-6 Astra on materials benchmark
giffmana · x · 2026-09-16
Liam Fedus (ex-OpenAI post-training lead) revealed his team built high-throughput materials labs in Menlo Park creating a loop between experiments and models: labs generate fresh data, models learn from it, then decide what to try next.
Using only 1,300 H200s plus months of proprietary experimental data for mid-training and RL, the resulting open-source model Neon surpasses GPT-6 Astra on their analysis benchmark — a showcase of small compute plus a tight experiment-data loop beating much larger models.
Related event: Periodic Labs beats GPT-6 on materials science with just 1,300 H200s(41 posts)→
More from AGI Musings
- Jensen Huang: 'the end of software' is nonsense, and AI destroying jobs is untrue — kimmonismus · 2026-09-16
- LLMs are unbeatable for functional work: devs care that scripts run, not how elegant they are — TivadarDanka · 2026-09-16
- "Everyone using AI will have a 5-7 year advantage over the rest of the population" — rand_longevity · 2026-09-16
- Polymarket gives Sam Altman just 5% odds of signing AI-slowdown statement — Polymarket · 2026-09-16
- "The worst people are asking to regulate AI" — AI circles debate regulation — wen_ragnarok · 2026-09-16
- US Drone Regulation Hollowed Out Its Industry: An AI Warning as DJI Took the Room — TinfoilTricorn · 2026-09-16