autonomous coding agentssoftware engineeringpublished Nov 11, 2026 · arXiv:2611.13246ExpertCrowd-SWE: Measuring Frontier Coding Agents on Expert-Authored Engineering Tasks in Unfamiliar Codebases
automated evaluationlanguage-model rankingpublished Aug 24, 2026 · arXiv:2608.20574FlavourBench: Ranking Frontier Language Models with Executable Culinary Ground Truth