Gemini 4 is almost ready, says new Google DeepMind chief

A faster, better Gemini 4 could unseat current AI leaders and boost Google's competitive standing.

Google DeepMind chief Koray Kavukcuoglu indicated that the Gemini 4 AI model is nearing completion and in its refinement stage. He stated the company aims for an "early post-training output" release, "as soon as possible," and certainly "much earlier" than year's end. This follows a period where Google has lagged behind competitors like OpenAI and Anthropic, who have released more advanced models such as GPT-6 and the Mythos series since Google's last flagship release, Gemini 3, in November 2025. A planned Gemini 3.5 Pro update in June was also not rolled out, with Kavukcuoglu explaining the company prioritized developing faster, less powerful "Flash" models instead. Despite perceived delays, Kavukcuoglu expressed confidence in Google's ability to remain at the forefront of AI development.

Google's impending Gemini 4 release signals a renewed push to regain market leadership in the generative AI race, a space where it has recently ceded ground to OpenAI and Anthropic. The focus on rapid iteration and early release of post-training outputs suggests a strategy to quickly incorporate user feedback and competitive pressures. The shift in focus from a large, delayed Gemini 3.5 Pro update to the development of "Flash" models also indicates a strategic pivot towards more agile and potentially cost-effective AI deployments.

This intensified development cycle under new leadership aims to challenge the performance benchmarks set by competitors and re-establish Google's position as an AI innovation leader. The success of Gemini 4 will be crucial in determining Google's ability to compete in the rapidly evolving AI landscape and influence future research and development priorities across the industry. Investors and developers will be watching closely to see if this release can close the perceived performance gap and restore Google's competitive edge.