Gemini 4 Is Here, and Google’s Flagship Tops All Other AI Models on Cybersecurity

What happened

In brief Google unveiled Gemini 4 Argon on Wednesday, scoring 77.9% on DeepSWE v1.1 and leading 12 of 18 benchmarks in its own comparison table. Google unveiled Gemini 4 Argon on Wednesday, calling it its frontier model, meaning its most capable, for coding, office work and cyber defense. Google is a software and Internet company based in Mountain View, and its products and services include Software tools.

Cyber defenders get it first, with the guardrails off. It posted a 0.7% attack success rate on Gray Swan's prompt injection test, ahead of Claude Opus 5.5 and Claude Fable 5.1, which both scored 1.0%.

Argon goes first to vetted cyber defenders through the Fairwind Program, without cyber guardrails (limits built in to stop a system doing certain things), before reaching paid API (the interface one piece of software uses to talk to another) customers and Google AI Ultra subscribers. Gemini 4 is finally here, one week after the release of Claude Opus 5.5 and one day after GPT 6.1 Sol, proving American labs are very much committed to slowing down AI development. On DeepSWE v1.1, a test of whether an AI can finish long, messy, real-world software engineering jobs, scored as a percentage, Argon hit 77.9%.