Google Unveils Gemini 4 Argon with Enhanced Security Features

Google has unveiled Gemini 4 Argon, its latest model for 'deep reasoning across complex, long-horizon workflows'. The company claims that this new model can handle tasks such as cybersecurity defense, coding, and creative writing with ease. One of the key features of Gemini 4 Argon is its ability to process up to 1 million tokens at a time, which is a significant increase from its predecessor's 64K token limit.

The model is said to be highly capable in cybersecurity defense, with Google training it to detect and prevent various types of attacks. Gemini 4 Argon will initially be available to Google AI Ultra subscribers and paid API customers at an introductory price of $2 per million input tokens and $10 per million output tokens.

Google is also taking steps to ensure the model's security, including monitoring its internal activations for misuse and hardening sandboxed environments. The company claims that Gemini 4 Argon is state-of-the-art in various benchmarking tests and is being used extensively by Google engineers internally.