Google's September 30 announcement of Gemini 4 Argon, its new frontier large language model, has shifted trader focus toward an imminent public rollout after months of delays. The model, optimized for long-horizon software engineering, cybersecurity, and complex knowledge work with a 1M-token output limit, is rolling out first to select cyber defenders via the Fairwind Program while undergoing safety evaluations. This phased approach follows leadership changes at DeepMind and competitive pressure from OpenAI's Astra and Anthropic's Opus, with Google highlighting benchmark gains in coding and agentic tasks. Broader availability to developers and subscribers is expected soon, though exact timing remains fluid amid ongoing testing.