Google Unveils Gemini 4 Argon: New Frontier Model Brings Deep Reasoning, More Complex Workflow

Google has unveiled Gemini 4 Argon, positioned as its most capable model yet for sustained professional work—especially software engineering, enterprise knowledge tasks, and cybersecurity defense. The initial release is limited: first to a select group of trusted cyber defenders via Google’s Fairwind Program, then expanding to paid API customers and Google AI Ultra subscribers. The model’s pricing begins at $2 per million input tokens and $10 per million output tokens, with cached input tokens 95% cheaper, rising after the introductory period to $4 and $20 respectively. Internally, Google says it has used Argon for optimizations, memory savings, and large-scale code migrations. A key technical feature is an expanded output context limit raised to 1 million tokens.