Google DeepMind unveils Gemini 4 Argon with 1M-token output for long workflows

Google DeepMind has introduced Gemini 4 Argon, a frontier model that can produce up to 1 million tokens in one response, versus 64,000 for earlier Gemini models. It targets long-running coding, legal and finance work, and cyber defense. Access starts with vetted security teams in the Fairwind Program, with paid API users next. Introductory pricing is $2 input and $10 output per million tokens.