From the source
- Google began rolling out Gemini 4 Argon, a frontier AI model, to vetted cybersecurity defenders in its Fairwind Program.
- The model outperforms
Anthropic's Claude Opus 5.5 and
OpenAI's GPT-6 Astra on most of Google's benchmarks, scoring 77.9% on DeepSWE v1.1 and 51.3% on AutomationBench.
- Argon has a 1 million token output limit and is priced at $2 per million input tokens and $10 per million output tokens at launch, with cached input 95% lower.