Google's Gemini 4 Argon is its smartest model yet, and for now it's strictly guest list only
By Brett Gavaghan, Founder of Adaptd ·
Google announced Gemini 4 Argon on 30 September, its new top model for long, complex work. Only vetted security teams can use it so far.
The short version
- Gemini 4 Argon is Google's new frontier AI model for long, complex work like coding, legal and finance tasks.
- It's limited to cyber defenders in Google's Fairwind Program, with paid API customers and Google AI Ultra subscribers next.
- Keep using your current Gemini tools and test Argon on real tasks when it reaches your plan.
Every few weeks an AI company tells us it has the smartest model in the world. This week it was Google's turn, with the small catch that most of us can't use it yet.
What is Gemini 4 Argon?
Gemini 4 Argon is Google's new top AI model, announced on 30 September and built for long, multi-step work. Google says it does well at software engineering, enterprise work like legal and finance tasks, and cyber defence.
It can write up to 1 million tokens in a single answer, and Google says thousands of its own staff already use it. On Google's own comparison it beats OpenAI and Anthropic's top models on most benchmarks, which is the kind of result every company seems to find about its own model.
Who can use it, and what does it cost?
Gemini 4 Argon is only available to trusted cyber defenders through Google's Fairwind Program for now. Google says a wider release comes as soon as possible, starting with paid API customers and Google AI Ultra subscribers.
For developers, the introductory price is US$2 per million input tokens and US$10 per million output tokens. After the introductory period it rises to US$4 and US$20. Google hasn't given a date for Workspace plans.
Should your business do anything yet?
Your business doesn't need to do anything about Gemini 4 Argon today, because it isn't on any standard Workspace or Gemini plan yet. Keep using the Gemini tools you have.
My take: when it arrives, give it a couple of your own hard tasks, like a messy contract summary, and compare it with what you use now. Benchmarks tell you who won the exam. Your own tasks tell you who'd be good at the job.
Sources
Wondering what this means for your business?
That's exactly what the discovery call is for.