Google announced Gemini 4 Argon on September 30, 2026, as its new frontier large language model optimized for long-horizon software engineering, enterprise tasks in legal and finance, and cybersecurity defense. The model debuted with an industry-leading 1 million token output limit and strong benchmark results, including 77.9% on DeepSWE v1.1, ahead of recent rivals like GPT-6 Astra and Claude Opus 5.5. It is currently rolling out in a phased manner through Google’s Fairwind Program to vetted cyber defenders and internal teams, with broader API access planned first for paid developers and Google AI Ultra subscribers. Pricing starts at an introductory $2 per million input tokens and $10 per million output tokens before rising. This limited early deployment and lack of a firm general availability date shape trader focus on the timeline for wider release.