A collection of test prompts run through 30+ local models, with performance notes, screenshots, and a rubric developed across multiple use cases. - View it on GitHub