Google said on September 30 it was rolling out Gemini 4 Argon, an AI model for coding, enterprise knowledge work, and cybersecurity defense, to trusted cyber defenders through its Fairwind Program. The company is testing safeguards before a wider release.

Google was also taking part in a voluntary US government process that gives officials pre-release access to AI models. It plans to expand Argon later to developers, enterprises, and consumers.

The disclosure came as the White House considered reviewing models before broad release, and Trump signed an order seeking early access to new AI systems on June 2.

Google priced Argon at US$2 per million input tokens and US$10 per million output tokens, with cached input tokens costing 95% less. It also raised the output limit to 1 million tokens from 64,000.

Google said Argon scored 77.9% on DeepSWE v1.1 for long-horizon software engineering tasks. It also tied for first place at 68% on CWE-bench v1, a benchmark for fixing security vulnerabilities.

Wiz, a New York-based cloud security company, is using Argon through its Scan for Good initiative, which Google said helped find a critical vulnerability in healthcare software.

The limited rollout reflects caution around cyber-capable AI. OpenAI has separately described trusted access for cyber defense tools.

🔗 Source: Google

Recent Google developments