Benchmarking Large Language Models for Root Cause Analysis in Train Control Software Testing

A benchmark of seven large language models for automated root cause analysis of test failures, using execution logs from Alstom’s digital twin-based testing of train control software.