because what we really needed was another benchmark for AI models to chase meaningless high scores in, said no one ever who actually has to solve real-world problems. https://modelrift.com/blog/openscad-llm-benchmark/