An AI system screened 2.4 million different crystals in just 28 hours and successfully guided the creation of four brand-new superconductors.
April 29, 2026
Original Paper
Agentic Fusion of Large Atomic and Language Models to Accelerate Materials Discovery
arXiv · 2604.23758
The Takeaway
Finding new materials usually takes years of trial and error in a laboratory. This agentic framework combines the reasoning of large language models with specialized physics models to predict material properties at light speed. The system didn't just guess, it provided the exact recipes needed for scientists to synthesize the materials. Discovering four superconductors in a single day is an unprecedented pace of progress in the field of materials science. This technology could accelerate the development of everything from better batteries to faster computers by decades. It represents a massive leap in how we discover the building blocks of the future.
From the abstract
The discovery of novel materials is critical for global energy and quantum technology transitions. While deep learning has fundamentally reshaped this landscape, existing predictive or generative models typically operate in isolation, lacking the autonomous orchestration required to execute the full discovery process. Here we present ElementsClaw, an agentic framework for materials discovery that synergizes Large Atomic Models (LAMs) with Large Language Models (LLMs). In response to varied human