Google Rolls Out Gemini 4 Argon to Paid Users and Developers

Google has opened up Gemini 4 Argon, its new flagship AI model, to paying subscribers and developers worldwide. As of October 13, Google AI Pro and Ultra subscribers can pick Argon in the Gemini app and in AI Mode in Search, and developers can call it through the Gemini API.

It's the moment Google has been building toward for months. After a rocky stretch of delays, the company finally has a broadly available model it says can compete with the best from OpenAI and Anthropic.

How we got here

Google announced Gemini 4 Argon on September 30, but only a small group could use it. Initial access went to vetted cybersecurity defenders through Google's Fairwind Program and, in the Gemini app, to top-tier Ultra subscribers.

The slow start followed a difficult year. Google told the public in September that it no longer planned to release Gemini 3.5 Pro, a model CEO Sundar Pichai had previously said would arrive in June. Reuters also reported that Google overhauled its DeepMind lab after co-founder Demis Hassabis stepped aside, and that several leaders of the Gemini model team departed.

Why this matters: Argon is Google's answer to the question of whether it can still ship at the frontier. A broad, priced release is a much stronger signal than a limited preview.

What's rolling out, and to whom

Here's the access breakdown:

• Gemini app and AI Mode: Google AI Pro and AI Ultra subscribers. Argon appears in the model picker alongside 3.8 Flash and other models.
• Antigravity (Google's agentic coding app) desktop and CLI: AI Ultra only.
• AI Studio and Android Studio Canary: AI Ultra only.
• Gemini API: paying developers and businesses worldwide.

API pricing starts at $2 per million input tokens and $10 per million output tokens until November 30, 2026. After that, prices double to $4 and $20. Cached input (text the model has already seen, like a long system prompt) gets a 95% discount.

What's different about Argon

Google says Argon "answers questions more directly with less hedging," repeats itself less, and avoids clichés. On longer jobs it gives more frequent progress updates, so you're not left staring at a spinner.

The app also gets a new toggle that switches between chat and tasks. In tasks mode, Gemini keeps working across Google apps while you're away. It connects to more than 40 apps, including Instacart, OpenTable, Adobe, and Shopify, so you can hand off things like grocery orders or restaurant bookings.

On macOS, you can now jump straight from chatting to coding inside the Gemini app using Antigravity's agentic coding interface.

Coding is the big pitch

Google is leaning hard on coding. It says teams using Argon in Antigravity increased agentic code submissions by more than 35% and cut code rollbacks by nearly a third.

The company also shared internal examples. Argon helped untangle a legacy folder of about 6 million lines inside Chromium's 40-million-line codebase, finishing months of work in weeks. On YouTube, it helped cut the CPU footprint of the comments frontend in half. These are Google's own numbers and haven't been independently verified.

Who this affects

Developers get a new top-tier option at an aggressive introductory price. If you're building agents or coding tools, the next seven weeks are the cheapest time to test it.

Enterprises get another credible vendor. Google named early testers including Deloitte, LexisNexis, Ambience Healthcare, and Isomorphic Labs, spanning consulting, legal, health, and drug discovery.

Security teams may get special access. Google is expanding Fairwind to hundreds more partners, giving vetted defenders a version of Argon without its cyber guardrails.

What the community is watching

Independent measurement matters here. Third-party benchmarker Artificial Analysis reports that Argon has the lowest hallucination rate (how often a model confidently makes things up) among top-scoring models, at 15%. That's a meaningful edge for business users who need answers they can trust.

Still, the leaderboard is crowded. Anthropic's Claude Opus 5.5 and OpenAI's GPT-6 Astra remain the benchmarks Argon will be measured against in real-world use.

What's next

Watch the November 30 price change. If Google extends the introductory pricing, that's a sign it's fighting hard for developer share.

Google is also offering 10,000 scientists six months of free access to Antigravity's highest usage tier. Expect early research results and case studies to surface over the coming months.

The bottom line

Gemini 4 Argon is Google's most important release of the year, and it's finally in people's hands. If you pay for Google AI Pro or Ultra, open the model picker and try it on a real task today. If you're a developer, test it before November 30, while it costs half as much.