Gemini

Google says Gemini 4 release is coming ‘as soon as possible’

Ben Schoon | Sep 24 2026 - 7:45 am PT

As its rivals continue to iterate quickly, Google has confirmed that its Gemini 4 flagship AI model is nearing release and that it is coming “as soon as possible.”

Google DeepMind chief Koray Kavukcuoglu confirmed to The Information at the publication’s AI Agenda Live Summit this week that its next flagship Gemini model, Gemini 4, has entered “post-training,” a phase in which the model’s behavior is refined ahead of a public release. This includes safety testing and implementing guardrails, with internal testing underway using Gemini 4 to power Antigravity.

In other words, Google is getting closer to the release of Gemini 4.

Kavukcuoglu explained:

Advertisement - scroll for more content

Our intention is to, like—as soon as possible—to release an early post-training output because we see the results and we are excited… [to] continue the fast-paced iterations.

He adds that he hopes the model will be released “much earlier” than the end of 2026, but a specific timeline was not shared.

The update comes just days after Anthropic and OpenAI have both launched major updates to their respective flagship AI models in recent weeks. Responding to the idea that Google has “fallen behind,” Kavukcuoglu said that “I have the utmost trust in the team” and that “in my mind, it’s a certainty that we are always gonna be at the frontier.”

Google’s last major flagship model release was in November 2025 with Gemini 3 Pro, followed by Gemini 3.1 Pro in February 2026. There have been several updates to Flash and other models in the time since, most recently with Gemini 3.8 Flash. Google teased Gemini 3.5 Pro over the Summer, but never ended up releasing the model, and this week’s conversation with The Information sheds no light on whether or not that model will be released – Kavukcuoglu said that Google “took a little bit of a step back” to focus on its Flash models instead, but that the current focus is on releasing Gemini 4.

Speaking on AGI, a common conversation around other major AI model updates, Kavukcuoglu says that “the conversation is more about are we able to build intelligent agents that we can trust.”