Periodic Labs Trains Materials Model Beating GPT-6 with Just 1,300 H200s

Periodic Labs, co-founded by former OpenAI post-training lead Liam Fedus, published a blog revealing how it trained its materials science model Neon: it built a high-throughput materials lab in Menlo Park, forming a closed loop of "lab produces new data → model learns → model guides next experiments," covering hard problems such as superconductors, magnets, and semiconductors. Using only 1,300 H200 GPUs plus months of proprietary experimental data, it mid-trained and RL-trained an open-weights model to Pareto-dominate GPT-6 Astra on X-ray diffraction benchmarks.

Confirmed

• Periodic published the blog "Building Labs that Learn," laying out its three pillars—high-throughput labs, research, and infrastructure—and arguing that data from frontier labs is being wasted while automated closed loops can let data continuously strengthen scientific AI
• Neon is a 1-trillion-parameter model already deployed in the company's own labs to guide experiments
• Peak compute was just 1,300 H200 GPUs (the team says H200), trained from open weights through mid-training and RL stages

Why it matters

• Demonstrates that the "proprietary experimental data + specialized closed loop" approach can use far less compute than general-purpose frontier models to beat models like GPT-6 Astra on specific scientific tasks, offering a replicable paradigm for vertical scientific AI
• Automated labs and models reinforcing each other could significantly accelerate the discovery of key materials such as superconductors, magnets, and semiconductors

2026-09-16 ~ 2026-09-16 · 31 related posts

• [source] Periodic Labs trains Neon on just 1,300 H200s to beat GPT-6 Astra on materials analysis — LiamFedus · 2026-09-16
• [source] Periodic Labs unveils "labs that learn": 1T-param Neon now deployed in its own labs — LiamFedus · 2026-09-16
• Periodic Neon lifts XRD analysis success from 2.7% to 55.3% — a 20x jump — LiamFedus · 2026-09-16
• Periodic Labs' Neon, trained on 1,300 H200s, beats GPT-6 Astra on its science benchmark — _AndrewZhao · 2026-09-16
• Periodic Labs trains Neon with just 1,300 H200s, beating GPT-6 Astra on materials analysis — LiamFedus · 2026-09-16
• Periodic Labs: 1,300 H200s beat frontier models on X-ray evals, 4.1x Megatron throughput — zephyr_z9 · 2026-09-16
• Liam Fedus' Lab Built Neon: 1,300 H200s and a Self-Driving Materials Loop Beat GPT-6 Astra at Superconductor Analysis — vwxyzjn · 2026-09-16
• Open-source model Neon mid-trained on 1,300 H200s claims to beat GPT-6 Astra — bookwormengr · 2026-09-16
• [source] Periodic Labs unveils 1T-parameter Neon model that beats GPT-6 Astra on diffraction analysis — agarwl_ · 2026-09-16
• Periodic Labs trains a 1T-parameter XRD analysis expert on its lab data — lishali88 · 2026-09-16
• OpenAI's Liam Fedus unveils Neon: scientific models trained with modest compute — UltraRareAF · 2026-09-16
• Periodic Labs midtrains and RLs a trillion-param LLM to analyze superconductor lab data, beating Astra and Fable — vwxyzjn · 2026-09-16
• OpenAI team trains Neon materials model with just 1,300 H200s, claims it beats GPT-6 Astra — willdepue · 2026-09-16
• Ex-OpenAI post-training lead Liam Fedus confirms move into physical-world AI — LiamFedus · 2026-09-16
• Neon: 1,300 H200s and automated labs beat GPT-6 Astra on materials benchmark — andrew_n_carr · 2026-09-16

16 near-duplicate retellings: vwxyzjn · LiamFedus · LiamFedus · vwxyzjn · LiamFedus · LiamFedus · LiamFedus · vwxyzjn · LiamFedus · vwxyzjn · LiamFedus · LiamFedus · sudoraohacker · BanghuaZ · ying11231 · LiamFedus