Google announced Gemini 4 Argon on September 30, 2026, as its new frontier large language model optimized for long-horizon software engineering, enterprise knowledge work in legal and finance, and cybersecurity defense. The model features a 1 million output token limit—up from 64,000 in prior versions—and leads or ties rivals like OpenAI’s GPT-6 Astra and Anthropic’s Claude Opus 5.5 on most of Google’s reported benchmarks, including DeepSWE v1.1 at 77.9 percent. It is currently rolling out only to vetted partners in the Fairwind Program and Google’s internal teams under a phased safety approach, with U.S. government pre-release review underway; broader API access for paid users and Google AI Ultra subscribers is described as “as soon as possible” once guardrails are validated, without a fixed timeline. Introductory API pricing of $2 per million input and $10 per million output tokens has also been disclosed.