ModelGemini

Google unveils Gemini 4 Argon for coding and security

Original title: Gemini 4 Argon: our next era of frontier intelligence

Plain language

What it is: Google announced Gemini 4 Argon, a new artificial intelligence model designed for software engineering, professional office work, and cybersecurity.
How it works: The model can handle complex, multi-step tasks and has a very large limit of 1 million output tokens, allowing it to think deeply and solve difficult problems in one go.
Where it is available: It is currently rolling out to a select group of cybersecurity defenders through the Fairwind Program, with plans to expand to developers, businesses, and consumers later.
Example: A cybersecurity company named Wiz used the model to find a critical security flaw in healthcare software that exposed sensitive personal information.

What it is: Google announced Gemini 4 Argon, a frontier model optimized for complex, long-horizon tasks in software engineering, enterprise workflows, and cybersecurity defense.
How it works: It features an expanded output token limit of 1 million tokens, up from 64K, enabling deep reasoning and multi-step problem solving. It also includes specialized safeguards against prompt injections and misalignment.
Where it is available: The model is initially available to trusted cyber defenders in the Fairwind Program, and will later roll out to paid API customers, Google AI Ultra subscribers, and the public.
Example: Google agents used the model to migrate C/C++ codebases to Rust, replacing 32K lines of SIMD code in the libgav1 library to create a memory-safe video decoder that runs 2.7x faster.

The headline and explanation on this page are based on Google's post. The full details are in the original.

Gemini 4 Argon: our next era of frontier intelligence