Google announced Gemini 4 Carbon on October 28 as its latest frontier large language model, claiming state-of-the-art results on cybersecurity benchmarks such as SEC-bench Pro and long-horizon agentic tasks while topping the updated Artificial Analysis Intelligence Index at 59. The company began limited access that week to vetted cyber defenders through the Fairwind Program under a deliberate rollout that includes U.S. government pre-release testing and guardrail iteration, mirroring the phased approach used for Gemini 4 Argon two weeks earlier. Traders see the November 9–15 window as the most probable for broader developer and consumer availability because Google has signaled expansion “as soon as possible” once initial feedback arrives, yet has set no firm timeline amid the model’s dual-use capabilities.