Google Delays Gemini 3.5 Pro Launch to July As It Tweaks Its New Frontier AI Model

My take: Gemini 3.5 Pro isn't coming in June: Google confirmed it's pushing it to July to fold in feedback from early testers and lessons learned with the Flash 3.5 model. The model promised a 2 million token context window (basically, how much information it can 'remember' at once) and extended reasoning, and it was already in internal review and a limited enterprise preview. The timing landed awkwardly for Google: the delay hits right as four key researchers, including Gemini co-lead Noam Shazeer, left this week for OpenAI and Anthropic. A technical delay on its own can be perfectly healthy (better they ship something polished than something half-baked). But on top of the talent drain, it does raise the question of how fast Google can keep sustaining its cutting-edge model development. Not a reason to sound the alarm, but it adds up. And in this race, losing your stride for a few months shows.