Inside Gemini 4 Argon: How Google's 1M-token flagship model changes enterprise AI

Google has officially rolled out Gemini 4 Argon, its next-generation flagship reasoning model designed to handle advanced, long-horizon professional workflows.

As per the tech giant, the new model model excels at "deep work" on complex, multi-step tasks in coding, finance, law, cybersecurity and science.

Built as a large-scale sparse mixture-of-experts (MoE) transformer, Argon excels across complex software engineering, enterprise legal and financial analysis, and cybersecurity defense.

A standout feature of the release is its unified 1-million-token context window, which supports long decodes enabling the model to think deeply and return complete multi-file code diffs, legal briefs, or technical proofs in a single pass.

Google emphasized that the model communicates more directly with less hedging and fewer clichés, marking a major leap forward in reliability and precision for autonomous enterprise AI agents.

The release follows a rocky stretch in Google's AI efforts. The company announced in September that it no longer planned to release Gemini 3.5 Pro, which CEO Sundar Pichai had originally said would be released in June.

It also overhauled DeepMind after the AI lab's co-founder and CEO Demis Hassabis stepped aside and several leaders of its Gemini model left the company.

Argon is available to paying customers through the company's AI application building tools. Google AI Pro and Ultra subscribers can access it through the Gemini app and AI Mode in Search, while the Antigravity coding tool is limited to Ultra subscribers.

API pricing at launch starts at $2 per million input tokens and $10 per million output tokens, with rates set to double after November 30.