Google's September 30, 2026 announcement of Gemini 4 Argon as its new frontier large language model drives current trader sentiment, with the model already rolling out in a controlled phase to vetted cybersecurity organizations via the Fairwind Program. The release emphasizes long-horizon reasoning, a 1-million-token output limit, and leading benchmark results in software engineering, legal/finance workflows, and cyber defense—outpacing recent rivals from OpenAI and Anthropic on Google's cited metrics—while pricing starts at $2 per million input tokens during an introductory period. Safety considerations prompted the phased approach, including U.S. government pre-release reviews, before broader access expands to paid API users and Google AI Ultra subscribers. Traders are watching for the next expansion milestone or any delays in public availability, as the model builds on internal Google deployments and follows the earlier skip of a Gemini 3.5 Pro variant.