DeepMind executive signals accelerated timeline for Google Gemini 4 launch

A senior DeepMind executive told attendees at The Information's AI Agenda Live Summit that Google's next-generation Gemini 4 AI model is in early post-training and could launch significantly earlier than the end of 2026. The statement underscores Google's effort to maintain competitive momentum in the rapidly evolving generative AI market, where Meta's newly launched Muse agent and ongoing efforts by OpenAI and Anthropic have intensified pressure on the company's AI strategy. CEO Sundar Pichai had previously announced in July that Gemini 4 would represent the company's most ambitious pretraining effort to date, though no specific timeline had been disclosed until now.

1. DeepMind Chief AI Architect Koray Kavukcuoglu said Gemini 4 is in early post-training with launch much earlier than year-end, at The Information's AI Summit on Sept 24.
2. CEO Sundar Pichai announced in July that Gemini 4 would be the company's most ambitious pretraining run, though no timeline was provided.
3. Meta's Muse AI agent launched in September and rapidly reached top of Apple's U.S. App Store, intensifying consumer AI competition.
4. JPMorgan analysts predicted Meta's Muse could become the most widely used consumer AI app since ChatGPT.
5. In May, Google's Gemini breached the systems of three companies during a cybersecurity test.