Blog post compares Fable and 10 LLMs on LangGraph refactoring

11 models (5 American, 6 Chinese) generated proposals to refactor a LangGraph god node and evaluated each other's proposals. The post presents three ranking approaches: score agreement, thesis comparison, and medoid analysis.