Open-source benchmark pits Kimi K3 against Claude models on code quality

A new open-source benchmark compares Kimi K3, Claude Fable 5, and Claude Opus 4.8 on real-world coding tasks, focusing on code structure and maintainability rather than leaderboard scores. The test uses a double-entry ledger module with hidden money-safety bugs that models must fix. Results show all models fixed atomicity bugs but varied in handling floating-point precision and internal accessor issues.