This ranking comes from a single Reddit user's subjective experience; the evaluation scenarios are heavily skewed toward role-playing and creative writing, involve a large number of de-censored fine-tuned models, and may not be applicable to general assistant or code generation needs.