Google releases Gemini 4 Argon AI model to cybersecurity defenders

Google commenced the rollout of its Gemini 4 Argon AI model to trusted cybersecurity defenders through its Fairwind Program on September 30. This new AI model is designed for coding, enterprise knowledge work, and cybersecurity defense applications.

The company is currently testing safeguards before a wider release. Google is also participating in a voluntary U.S. government process that provides officials with pre-release access to AI models. The company plans to expand access to Argon later for developers, enterprises, and consumers.

Gemini 4 Argon is priced at $2 per million input tokens and $10 per million output tokens, with cached input tokens costing 95% less. The output limit has been increased from 64,000 to 1 million tokens.

In performance benchmarks, Argon scored 77.9% on DeepSWE v1.1 for long-horizon software engineering tasks. It also achieved a 68% score on CWE-bench v1, a benchmark for fixing security vulnerabilities. Wiz, a cloud security company based in New York, is using Argon through its Scan for Good initiative, which Google stated helped identify a critical vulnerability in healthcare software.