We recently conducted an external evaluation comparing ChatGPT, Gemini, Claude and Perplexity on a real-world hotel recommendation task in Greater London. The study used a pre-existing deterministic benchmark covering 167 eligible five-star hotels and 148…