Sources: some Google employees say Gemini 4 performs well on benchmarks but struggles with some real-world coding tasks; Google disputes that characterization (Bloomberg)

Google’s upcoming Gemini 4 launch highlights a common enterprise AI risk: strong benchmark scores may not translate into reliable performance on real-world coding and development tasks. For CIOs and technology leaders, the key implication is that vendor claims should be tested against actual enterprise workflows, since productivity, quality, and developer trust depend more on practical accuracy than on synthetic benchmark wins.