DeepMind's New Chief Says Gemini 4 Nears Debut, Signaling Google’s Push to Close AI Gap

The DeepMind arm of Google is set to roll out Gemini 4, its newest AI language model, following a stretch of relative silence while competitors have launched flagship systems in recent months.

In his inaugural public interview with The Information, new DeepMind chief Koray Kavukcuoglu said the model is "almost ready" for launch. His remark provides the first solid indication from the freshly installed head that Google intends to narrow the distance that has emerged between its Gemini line and rivals like OpenAI’s GPT‑4 and Anthropic’s Claude.

Gemini 4 succeeds the prior Gemini 1‑3 series, which Google released as a component of its wider plan to embed cutting‑edge generative AI throughout its offerings, from search to Workspace. Although those earlier models delivered respectable results, they have been outshone in public opinion by more heavily talked‑about rivals, creating internal demand to speed up progress.

Analysts point out that DeepMind’s research heritage—rooted in advances such as reinforcement learning and protein folding—has historically leaned toward pure research rather than productization. Kavukcuoglu’s hiring and his prompt mention of an imminent launch imply a pivot toward market‑ready offerings, bringing DeepMind’s work in line with Google’s profit‑centered AI goals.

The schedule for Gemini 4’s introduction may have notable implications for the wider AI landscape. As businesses and developers assess large language models for adoption, a fresh Google entry would heighten competitive pressure, possibly affecting pricing, feature configurations, and the overall speed of innovation in the field.

Although Google has not disclosed a formal launch date, analysts anticipate that limited beta trials will start within weeks, with a wider release slated for later in the year. The firm is expected to tap its vast cloud infrastructure to scale the system, making Gemini 4 a key pillar of its AI‑first strategy.