OpenAI team trains Neon materials model with just 1,300 H200s, claims it beats GPT-6 Astra
willdepue · x · 2026-09-16
William Depue of OpenAI detailed his team's automated materials labs in Menlo Park, which close the loop between experiments and models: labs generate fresh data, models learn from it, and then decide what to try next.
- Using only 1,300 H200s plus months of proprietary experimental data, the team mid-trained and RL'd an open-source model into "Neon"
- Neon reportedly surpasses GPT-6 Astra on the team's internal analysis benchmark
- OpenAI post-training lead Liam Fedus publicly confirmed and praised the effort
A landmark AI-for-Science case of experiment-in-the-loop training with modest compute.
More from Models
- 'You Can Feel the Gradient Slopes': Dev Contrasts Benchmaxxed RL Models vs General Ones — theteknosaur · 2026-11-02
- LightOnOCR-3 Launches: Tops FRBench, Adds Layout Structure and Chart Extraction — IgorCarron · 2026-11-02
- NVIDIA's Kumo Is a Family of Four Tabular Foundation Models, Not Just One — pandeyparul · 2026-11-02
- SemiAnalysis: safety classifiers are underrated — Meta's Muse tops cyber benchmark at 35.2% — ThomasScialom · 2026-11-02
- DeepSeek V4.1 Flash 1017 wins over a longtime closed-source-only user: fast, cheap, reliable — _bachrc · 2026-11-02
- Local quants vs Claude on 9 real private bugs: 3-bit costs thinking time, not score — FeydRowan · 2026-11-02