A research lab developed Periodic Neon, an AI model that analyzes X-ray diffraction data for materials discovery, achieving 55.3% success on complex internal evaluations while outperforming GPT-6 Astra and Claude Fable 5.1 at lower cost. The model automates hours of scientific analysis work by reasoning about synthesis conditions, crystal structures, and experimental context to identify material phases in powder samples.
Periodic Neon is an AI model trained to analyze X-ray diffraction (XRD) data for materials discovery, achieving a 55.3% success rate on complex internal evaluations—a 20x improvement over baseline models—while outperforming GPT-6 Astra and Claude Fable 5.1 at lower cost. The model automates the time-consuming task of identifying crystal phases and proportions in multiphase powder samples, freeing scientists to oversee more experiments in autonomous labs.
Periodic Neon, an AI model trained on lab data, achieves 55.3% success on complex X-ray diffraction analysis—a 20x improvement over baseline models—and is now deployed to automate XRD interpretation for materials discovery, freeing scientists from hours of manual analysis.
Periodic Labs has built high-throughput experimental facilities in Menlo Park to discover new materials using AI-guided scientific processes. The company developed Periodic Neon, a trillion-parameter AI model trained on lab data that outperforms GPT-6 Astra on materials analysis tasks like X-ray diffraction interpretation. They employ a cyclical approach combining hypothesis generation, synthesis prediction, and material characterization to enable increasingly autonomous scientific discovery.