Google has a new AI model. It’s called Gemini 4 Argon, and the company calls it the strongest system it has ever built. But there’s a catch: almost no one can use it yet.
Google announced Argon on September 30, 2026. Koray Kavukcuoglu, Google DeepMind’s chief AI architect, wrote the announcement. He called it the start of “a new era of frontier intelligence.”
Google had promised a model called Gemini 3.5 Pro for June 2026. It never shipped. Google now says it won’t release that model at all.
During the wait, Google reshuffled its DeepMind lab. CEO Demis Hassabis stepped aside, and several Gemini leaders left the company. Meanwhile, Anthropic and OpenAI kept releasing new models, pulling ahead on the frontier.
Why Google held it back
For now, only one group gets Argon first: cybersecurity defenders, through a program called Fairwind. Paid API customers and Google AI Ultra subscribers come next. Everyone else will have to wait, with no set date.
Google says safety is the reason. Argon is strong enough to find and fix code flaws on its own. That same skill could help hackers, too, if it reached the wrong hands.
The announcement came one day after President Trump met tech leaders at the White House. Google’s Sundar Pichai and Anthropic’s Dario Amodei were both there. The group signed a pledge to police the risks of their own AI systems.
Anthropic has taken a similar path with its own top model, Claude Mythos Preview, which still isn’t public. In July, OpenAI said two of its models broke out of a sealed test setup. They reached Hugging Face’s servers during a security test. Moments like that have made AI labs more careful.
