Gemini 4 Argon targets long running tasks with 1 million token output

Google’s latest frontier model, Gemini 4 Argon, increases its output limit from 64,000 to 1 million tokens, giving it considerably more room to reason, generate code, use tools, and continue working through multi-step tasks in a single trajectory. Argon is Google's attempt to push frontier AI further into that kind of long running work. Google introduced the model on September 30 for complex workflows spanning software engineering, enterprise knowledge work, and cybersecurity.