LLMatDesign presents a novel approach to materials discovery using Large Language Models (LLMs) for rapid screening and generation of materials. The framework interprets human instructions, applies modifications, and evaluates outcomes, demonstrating adaptability in a zero-shot manner. It showcases the potential for autonomous materials discovery in computational settings.

• Interpretable and efficient materials design
• Rapid adaptation to new tasks
• Successful development of materials with user-defined properties
• Autonomous decision-making in materials discovery

The framework highlights the capabilities of LLMs in advancing materials discovery processes and sets the stage for self-driving material laboratories.