IndustryCode: Benchmark Exposes LLM Code Generation Gaps in Finance, Automation, and Aerospace

arXiv·medium signal

New benchmark targeting industry-specific code generation in finance, automation, and aerospace — domains where existing code benchmarks (HumanEval, MBPP, SWE-bench) provide little coverage. Tests LLMs on the domain-specific constraints, regulatory compliance, and performance requirements that distinguish industrial code from general-purpose programming tasks.