Google announced Gemini 4 Argon on September 30, 2026, positioning it as the company's most advanced frontier large language model with state-of-the-art results on long-horizon coding, enterprise knowledge work, and cybersecurity benchmarks, directly competing with OpenAI's Astra and Anthropic's Opus. The model entered limited rollout exclusively to vetted cybersecurity partners through the Fairwind Program for vulnerability testing and guardrail refinement, with no public or developer access yet despite earlier signals of an October target and "as soon as possible" broader availability. This phased approach follows months of delays that caused Google to skip the planned Gemini 3.5 Pro, amid internal leadership changes at DeepMind and competitive pressure from faster-releasing rivals. Traders are watching for official updates on enterprise API access or full consumer release timelines, which could shift implied probabilities around year-end resolution criteria.