The new Gemini series benchmarks show a focus on speed and token efficiency, rather than significant intelligence improvement, leading some to interpret this as a slowdown in Google's AI pace compared to rivals. Google's strategy aims for mass adoption by optimizing for low cost and high speed for routine tasks, which constitute 95% of future use cases.