What Happened

Google announced Gemini 4 Argon, a new flagship AI model initially released only to select cybersecurity partners through its Fairwind Program. The model is designed for complex, long-horizon tasks across coding, enterprise knowledge work, and cybersecurity defense, and can autonomously find, validate, and patch critical software vulnerabilities.

Google priced Argon at $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached input tokens. The model's output token limit has been raised to 1 million tokens, up from a previous cap of 64,000.

Who Is Affected

Cybersecurity partners like Wiz are among the first deployers, using Argon through the Scan for Good initiative to provide free security scans of critical public infrastructure. Google said it will make the model available to trusted partners without standard cyber guardrails so defenders can use its full capabilities.

Paid API customers and Google AI Ultra subscribers will gain access once safety evaluations are complete, but no public release timeline has been set. Security ISVs and AI infrastructure vendors should watch for partnership and integration opportunities in the interim.

Market Impact

Argon ranks first on the Vals Index and ties for first on CWE-bench v1 with a score of 68%, and ranks first on AutomationBench with 51.3%. However, it finished behind rivals on two of four coding benchmarks cited in Google's own release documentation.

The pricing and token limits create immediate needs for cost management and observability tools. Internal results also show Argon freed more than 300 terabytes of memory across Google's data centers and beat a published quantum computing baseline by 40%.

What to Watch

Google has joined the Trump administration's voluntary pre-release review process and declined to say when Argon will become publicly available. The company plans to expand access to paid API customers and Google AI Ultra subscribers once safety evaluations are complete.

Watch for whether Google can avoid further delays after scrapping internal candidates for a Gemini 3.5 Pro model and confirming it no longer plans to release it. Leadership changes at Google DeepMind, including Demis Hassabis stepping aside as CEO, add further uncertainty to the roadmap.

Related coverage

• Worldline Launches Payment Tool For Purchases Made By AI Agents
• OpenAI begins rolling out Astra model after warning of its advanced cyber capabilities
• JFrog Artifactory vulnerabilities actively exploited for rapid admin access
• OpenAI launches 'Dots' always-on AI agents and 20+ announcements at DevDay