Google announced Gemini 4 Argon on September 30, 2026, marking its first frontier large language model release in months and positioning it competitively against OpenAI’s Astra and Anthropic’s Opus on coding, cybersecurity, and enterprise knowledge-work benchmarks. The model entered limited rollout to trusted cyber defenders via the Fairwind Program, with a 1M-token output limit and internal use already optimizing Google data centers, following leadership changes at DeepMind and the decision to skip Gemini 3.5 Pro. Traders are watching the phased safety review and voluntary pre-release process with U.S. authorities, which precede planned expansion to paid API users and broader availability. These steps, combined with recent post-training updates from the new DeepMind chief, shape near-term market sentiment around public access timelines.