Google Unveils Gemini 4 Argon AI Model for Cybersecurity and Coding Tasks

Google has launched Gemini 4 Argon, an artificial intelligence model designed for coding, research, and writing. The company claims it is particularly suited for cybersecurity work, where it can 'autonomously find, validate, and patch critical software vulnerabilities.'

Argon is being rolled out to a select group of Google's cyber partners through its Fairwind Program, which focuses on security initiatives. Google employees have already used the model in daily work, including debugging and codebase migrations.

The company says Argon also excels at coding and engineering tasks, as well as parsing visuals such as long videos and charts. It was 'built to sustain deep reasoning across complex, long-horizon workflows.'

Google claims that Argon outperformed OpenAI's GPT-6 Astra and Anthropic's Fable and Opus models in multiple AI benchmarks. Vals, an AI benchmarking startup, also showed that Argon leads its model index.