Google Delays Flagship Gemini 3.5 Pro as Coding Race Heats Up

On July 17, 2026, Bloomberg reported that Google has postponed the launch of its flagship model, Gemini 3.5 Pro, by several months. The delay comes because the model's coding capability has not yet reached Google's internal targets. Two weeks ago, the company refreshed its training data in an attempt to close the gap with rivals, but the results were disappointing, and executives decided to give the team more time to improve. The postponement is significant because AI code generation has become the most important battleground in the model race. OpenAI and Meta have already released systems that score higher than Google's current flagship on widely used coding benchmarks. Inside Google, some researchers fear that the company is no longer the obvious leader it was two years ago, and that a long delay could give competitors more time to lock in enterprise customers. The news has hurt morale inside the AI unit. According to ten current and former employees, many engineers and scientists are frustrated, and a steady talent drain toward OpenAI and Anthropic has made the situation worse. Several executives have asked leadership to take bolder steps, while co-founder Sergey Brin is said to be pushing for a more focused coding effort. Shareholders reacted quickly, pushing Alphabet's stock down more than 4 percent on the day of the report. Google has publicly said it is moving fast on multiple model families at once, but the episode shows that even the largest AI labs can stumble when the bar keeps rising. The next test of Gemini 3.5 Pro is now expected later this year, and the industry will be watching closely.