Recent claims that Google's Gemini AI outperformed Anthropic's Claude in 'Pokémon' games have sparked debate over AI benchmarking. While Gemini reached Lavender Town, it utilized a custom mini map, presenting an unfair advantage. In contrast, Claude struggled with raw image decoding, underscoring the challenges in creating reliable benchmarks. Critics argue that such tailored testing environments distort true AI performance, complicating comparisons. This controversy calls for standardized benchmarking methods in the industry to ensure transparency and trust among consumers and researchers.