Ever asked an AI a simple question and gotten three paragraphs of “great question!” before the answer? Google says it’s fixing that. On 13 October it started rolling out Gemini 4 Argon, its new flagship model, and one of the headline promises is refreshingly human: less fluff.
There’s a lot more going on, though. Argon can now work on tasks while you’re away, it plugs into dozens of apps, and its pricing is aimed squarely at developers. So let’s walk through what changed, what’s hype, and whether you should care.
So what actually changed this week?
Gemini 4 Argon first went to a small group of vetted cyber defenders through Google’s Fairwind Program. Then, on 13 October, Google opened it up much wider. According to Google’s own announcement, it’s rolling out globally across the Gemini app, AI Mode in Search and the Gemini API.
But there’s a catch. In the app, you need a paid Google AI Pro or Ultra plan. Engadget puts those at $20 a month for Pro and at least $100 a month for Ultra in the US. Free users and Google AI Plus subscribers don’t get it yet, though Google says “wider availability could follow later.”
Developers can call the model through AI Studio and the Gemini platform. Ultra subscribers also get it inside Antigravity, Google’s agentic coding tool. In plain terms, an agentic tool is software that can take steps on its own instead of only answering questions. Our AI agent explainer goes deeper.
Less fluff, more getting stuff done
Google says Argon answers more directly, with less hedging, less repetition and fewer stock phrases. It also claims the model “follows style instructions better than any of our previous Gemini models.” If you’ve ever begged a chatbot to stop using bullet points, you’ll get why that matters.
Then there’s the bigger shift. The Gemini app now has a tasks toggle in the main chat window. Flip it, and Gemini keeps working across Google apps while you’re away from your computer. For longer jobs, Google says, “it keeps you in the loop by sharing its progress more often and more effectively.”
Google also says “Gemini now connects you to more than 40 apps,” naming Instacart, OpenTable, Adobe and Shopify. In one demo, Gemini built a client proposal deck from a project folder, emailed it, and then set up a scheduled workflow for future requests. That’s an assistant that does the work instead of describing it. If it works reliably, it’s a real change.
Why developers care about Gemini 4 Argon
Here’s where it gets interesting for builders. Argon’s API costs $2 per million input tokens and $10 per million output tokens at its introductory price. A token is a small chunk of text, roughly a piece of a word. After 30 November, those prices double to $4 and $20.
The model also has a 1 million token limit shared between input and output. In other words, you can feed it very long documents or codebases in one go. We’ve explained how context windows work if that number feels abstract.
Benchmarking firm Artificial Analysis says Argon matches OpenAI’s GPT-6 Astra on its Intelligence Index, Engadget reports. It also does so at 60 percent of the cost per task, at introductory prices. Google adds its own stats, too. Inside Google, agentic code submissions rose by more than 35%, while code rollbacks fell by nearly a third.
One example stands out. Google says Argon helped split core webpage handling out of a legacy folder of about 6 million lines inside Chromium’s 40 million-line codebase. Work that normally takes months took weeks, the company claims.
Is it as good as Google says?
Honestly, we don’t know yet, and neither does anyone outside Google. Most of the big performance claims come from Google itself. That includes its claim that Argon ranks first on the Vals Index, which tests finance, coding, legal and tax work.
The rollout also looks bumpy. Android Police writer Sanuj Bhatia put it simply: “I’m subscribed to Google AI Pro myself, and I’m still not seeing Gemini 4 Argon available in the app just yet.” So if you pay and don’t see it, you’re not alone.
There’s some backstory as well. Google had planned a 3.5 Pro model, Engadget notes, but it moved on and released Argon instead. That tells you how fast this race is moving. Koray Kavukcuoglu, Google’s Chief AI Architect, calls Argon “just the beginning,” with the rest of the Gemini 4 family still to come.
One thing we did like: Google says Argon is trained to pause for human confirmation before irreversible steps. For an AI that now runs tasks while you’re away, that’s exactly the guardrail you’d want.
Okay, but should you care about Gemini 4 Argon?
If you already pay for Google AI Pro, yes. You’re getting Google’s newest model inside a plan you already pay for, so it’s worth testing on your real work. Try a long report, a messy spreadsheet or a coding task you’ve been putting off.
If you’re a developer, the pricing deserves a close look. A cheaper model that keeps up with the top tier changes the maths for startups. That’s especially true for small teams in Armenia and other emerging markets, where every dollar of API spend counts.
And if you’re on the free plan? Then nothing changes for you today. Still, keep an eye on it. Features that start behind a paywall often spread later, and Google has hinted as much.
The bigger picture is a price fight. We recently covered how ChatGPT started answering with buttons and maps. Now Google is pushing hard on agents and cost. That competition is good news for users.
What we’d actually do this week
First, don’t take anyone’s benchmark chart at face value, including Google’s. Instead, run Gemini 4 Argon on three tasks you actually do every week and compare it with whatever you use now.
Next, if you’re building a product, do the cost maths before 30 November, because that’s when the introductory price ends. Finally, if you try the tasks feature, start small. Let it book a table before you let it email a client.
Here’s our open question: if your AI assistant can quietly work while you’re offline, what’s the first job you’d actually trust it with?