We are excited to unveil our latest model, Gemini 4 Argon, which will be distributed to select cybersecurity defenders as part of our Fairwind Program. Designed to facilitate in-depth reasoning for intricate, long-term tasks, Argon is set to revolutionize how we operate and innovate at Google. This model offers exceptional performance in various complex tasks, including real-world software engineering, enterprise-level knowledge activities such as legal and financial services, as well as cybersecurity solutions.
To ensure a safe rollout of such advanced capabilities, we are adopting a gradual, phased strategy. We are currently collaborating with the U.S. government's voluntary initiative for pre-release model access and will broaden access over time. We are committed to collecting insights from initial testers as we refine safety measures before making Argon widely available to developers, businesses, and consumers as soon as possible.
At launch, Argon will be available at an introductory rate of $2 for every million input tokens and $10 for every million output tokens, with cached input tokens offered at an impressive 95% discount from the input token price.
