Gemini 4 Argon: Built to solve your hardest problems

Google has unveiled Gemini 4 Argon, the company's latest flagship model, to developers through paid access on the Gemini API and within various tools like Google AI Studio and Firebase AI Logic. Launched on September 30, the model initially offered limited access only to selected cybersecurity defenders in Google's Fairwind Program, trusted testers, and internal Google employees.

Priced competitively at $2 per million input tokens and $10 per million output tokens for the first month, the cost will increase to $4 and $20 per million tokens afterward, aligning with prices from OpenAI's GPT-6.1 Sol and Anthropic's Claude Opus 5.5. One key feature is its unified 1 million input/output token limit, which combines both input and output, unlike previous Gemini models with separate caps.

In benchmark testing, Argon scored first on the Vals Index in areas like finance, coding, legal, and tax work, outperforming competitors like Claude Opus 5.5, GPT-6 Astra, and Claude Fable 5.1. In embodied reasoning tasks, Argon scored 87 percent, surpassing its competitors. Google has also integrated Argon into its Antigravity platform for coding agents and the Gemini app, expanding its availability to more developers and enhancing the user experience.