A specialist that does one thing: judge whether two pieces of text are duplicates. It outputs a score rather than text, making it a silent scorer rather than a conversational partner. Its 514-token context window is tight, so it works best with short, focused inputs like questions or brief passages.