Kimi K3 leads the web engineering benchmark, ahead of Fable

The open-source Kimi K3 model tops the web engineering benchmark comparison in question, outperforming the Fable model and achieving similar success in less time. This is the first time an open model has outperformed every proprietary participant in such a comprehensive test; peak performance is 92% and 96% with additional assistance. The measurement may be an important signal of the advance of open models, although benchmarks alone do not convey the whole picture.