Fedus team builds material labs, trains open Neon on 1,300 H200s past GPT-6 Astra
eliebakouch · x · 2026-09-16
Liam Fedus' team built high-throughput materials labs in Menlo Park creating an experiments-models loop: labs generate fresh data, models learn and suggest next experiments. Using only 1,300 H200s plus months of proprietary experimental data, they mid-trained and RL'd an open-source model, Neon, surpassing GPT-6 Astra on their analysis benchmark. Elie Bakouch asks how much the periodic harness improves over Claude Code and Codex.
More from Models
- LightOnOCR-3 Reads 16th-Century Manuscripts and Ancient Greek Without Training — IgorCarron · 2026-11-02
- '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