Google Delays 'Gemini 3.5 Pro' Launch Amid Performance Shortfalls and Coding Limitations

Google's upcoming flagship AI model, 'Gemini 3.5 Pro,' has faced multiple-month delays due to unmet performance benchmarks in key areas like coding capabilities. Internal concerns suggest the delay risks ceding leadership to competitors OpenAI and Anthropic. Bloomberg reports engineers and executives are expressing frustration over the postponement. Despite recent updates to Gemini's training data, results fell short of internal expectations. With OpenAI and Meta recently surpassing Google in coding capabilities, the delay has intensified internal pressures. Google's stock briefly dropped 3.2% following the news. A spokesperson emphasized efforts to maintain cost efficiency while deploying models rapidly, noting ongoing tests with partners and discussions with the U.S. government on regulatory frameworks. Sundar Pichai had previously announced a June launch but pushed it to July, with rumors of an early July release later delayed. Insiders attribute the delay to Google's bureaucratic complexities and overlapping departmental interests. The company's fragmented approach, with separate teams developing AI coding tools, has hindered cross-functional synergy. A former employee likened the coordination challenges to 'boiling the ocean.' Cultural conflicts among engineers, particularly between strict adherents to manual coding standards and others, also contributed to the delay.