Overview
- Google announced Gemini 4 Argon on Wednesday, September 30, 2026, as a new frontier model built for long, multi‑step tasks in software engineering, enterprise knowledge work and cybersecurity.
- Argon raises the model's maximum single‑run output from 64,000 tokens to 1 million tokens so it can sustain extended reasoning and generate very large results in one session.
- Initial access is limited to vetted cyber defenders through Google’s Fairwind Program and to the U.S. government's voluntary pre‑release process, with paid API customers and Google AI Ultra subscribers scheduled to follow.
- Google-published benchmarks show Argon leading many enterprise and knowledge‑work tests, and the company says the model can autonomously find, validate and patch software vulnerabilities, a capability Wiz has already used to surface a critical healthcare flaw.
- Introductory API pricing is $2 per million input tokens and $10 per million output tokens with higher rates after the introductory period, and Google says it will expand access only after gathering tester feedback and completing further safety work.