Periodic Labs' Neon, trained on 1,300 H200s, beats GPT-6 Astra on materials analysis
LiamFedus · x · 2026-09-16
Liam Fedus unveiled Periodic Labs' Neon: the company built high-throughput materials labs in Menlo Park to close the loop between experiments and models, with labs generating fresh data that models learn from to pick the next experiments. Using only 1,300 H200s plus months of proprietary experimental data, they mid-trained and RL'd an open-source model to surpass GPT-6 Astra on their analysis benchmark. Initial focus: hard materials science problems including superconductors, magnets, and semiconductors.
Related event: Periodic Labs beats GPT-6 on materials science with just 1,300 H200s(41 posts)→
More from Companies & People
- Jensen Huang: 'the end of software' is nonsense, and AI destroying jobs is untrue — kimmonismus · 2026-09-16
- Chris Lattner to keynote PyTorch Conference on Mojo and MAX for heterogeneous compute — PyTorch · 2026-09-16
- Flipbook team Common Knowledge joins Figma to build the future of creating with AI — zan2434 · 2026-09-16
- Jensen Huang: 'Safety is an engineering problem' — every company will become an AI company — kimmonismus · 2026-09-16
- Stanford CS329A publishes 9-lecture course on self-improving AI agents — le_james94 · 2026-09-16
- Dario Amodei: no backing away from AI slow-down, calls for common standards — kimmonismus · 2026-09-16