Researchers unveil Regression Language Models (RLMs) that accurately predict numeric outcomes of code execution—like memory usage and latency—across programming languages without running programs. This breakthrough could revolutionize performance optimization and hardware-aware AI development by providing instant insights during coding.