Google delays Gemini 3.5 Pro to improve coding

Google has delayed the release of its flagship Gemini 3.5 Pro artificial intelligence model, missing a planned June launch as it works to improve the model's coding capabilities amid intense industry competition. The delay, first reported in mid-May, has created internal frustration and contributed to a decline in Alphabet's stock price.

The company had announced Gemini 3.5 Flash at its I/O developer conference in May and indicated the more powerful Pro version would follow in June. That deadline passed without an update. According to internal reports, Google updated the model's training data in late June specifically to enhance coding performance, but the results were disappointing, leading to a reset in development. The company is now testing Gemini 3.5 Pro, an upgraded Flash model, and other systems with partners, and is also discussing model capabilities and testing standards with the U.S. government.

The postponement comes as rivals like OpenAI and Meta have released models that outperform Google's current offerings on coding tasks. Internally, some researchers view access to Google Search data as Gemini's key advantage, while acknowledging competitors lead in overall model...