Model Forensics evaluates non-capability personality differences between models: style, tone, and character. For users these differences matter, nobody wants to use an annoying model.