Google released Gemini 4 Argon to security teams first
The October 2026 model can find and repair software flaws on its own, so Google limited early access to vetted defenders.
Kivorent News desk, published , 1 min read
68%Gemini 4 Argon's reported score on a vulnerability benchmark
This is a look back at an event from . We wrote it on November 11, 2026, from the sources listed at the end.
Google released a model called Gemini 4 Argon on October 1, 2026, and gave it first to vetted security professionals through a program it calls Fairwind. Google said the model scored 68% on a benchmark of software weaknesses and can discover and repair vulnerabilities without help, according to AI Weekly's summary of the announcement.
Why it mattered
A model that can find flaws in software is useful to the people who fix them and to the people who exploit them. Releasing it to defenders first gives them a head start.
It was a change in how AI models are launched. For most of the industry's history the newest model went to everyone at once. The effects were already visible: Apple's July security update patched hundreds of flaws in one release.
Sources
- AI news for October 1, 2026, AI Weekly
GoogleAI modelsInternet, outages and security
Spotted an error? Tell us and we will fix it.