Google Opens Up Gemini 4 Argon — But Only If You Pay

Google Opens Up Gemini 4 Argon — But Only If You Pay

After a Three-Week Wait, Argon Goes Global

Google released Gemini 4 Argon worldwide on October 13, 2026, opening the flagship it first unveiled on September 30 to a much wider audience. Access now runs through the Gemini API, AI Studio, Google's Agent Platform console and the Antigravity agentic development environment, while consumers get it inside the Gemini app and AI Mode in Search. The announcement came in a blog post from Koray Kavukcuoglu, the DeepMind executive who serves as the lab's chief AI architect. For the previous two and a half weeks, Argon had been confined to Google's internal teams and a hand-picked group of cybersecurity defenders working through the Fairwind Program.

Google Calls It Its Strongest Model Yet

The official pitch clusters around three jobs: coding, cybersecurity defense, and professional knowledge work in finance and law. Reuters quoted Google saying the model excels at "deep work" on complex, multi-step tasks across coding, finance, law, cybersecurity and science. Tulsee Doshi, Gemini's product lead, called Argon a "powerhouse" and said it is the first of a family of new Gemini models arriving over time. The benchmark numbers Google published include 77.9% on DeepSWE v1.1, a record for long-horizon software engineering; a tie for first at 68% on CWE-bench, which measures automated vulnerability patching; and 68.9% on the Vals Index, where Google also claims a lead on a finance-and-law evaluation. The biggest spec change is output: a single response can now run to 1 million tokens, up from 64,000. On safety, Quartz reported that Google says new anti-jailbreak measures cut violations on an adversarial test set by close to 50% versus earlier Gemini models.

The Introductory Price Lasts Six Weeks

Pricing is the sharpest edge of this launch. Argon enters the API at $2 per million input tokens and $10 per million output tokens, with cached input discounted 95%. That rate holds until November 30, 2026, then doubles to $4 and $20. The Verge noted the launch price is roughly a fifth of what OpenAI's GPT-6 Astra and Anthropic's Fable 5 cost. Ars Technica cautioned developers that the discount window is short and anyone building workflows around Argon needs to price in the increase now.

A Delayed Flagship Under Pressure

Reuters framed the release as Google trying to regain ground against OpenAI and Anthropic after months of delays and a leadership shakeup at DeepMind. The same day, Bloomberg reported that employees with direct access to the model said it struggles on real-world coding tasks, particularly front-end design — the work that decides how an app or website looks and feel. One employee said there is "large consensus" internally that Gemini 4 is frontier-class. Google pushed back, telling Bloomberg it would be "inaccurate" to say Gemini 4 underperforms at coding, and Kavukcuoglu has publicly said "it's a certainty that we are always gonna be at the frontier." Outside commentators have invoked "benchmaxxing" to describe the gap between leaderboard scores and day-to-day results.

Paying Subscribers Get It First

The rollout is tiered. The Gemini app and AI Mode serve Google AI Pro subscribers at $19.99 a month in the US and AI Ultra at $99.99 and up; Antigravity, AI Studio and Android Studio Canary require Ultra or a paid API key. The Verge and Ars Technica both reported that it remains unclear when — or whether — Plus subscribers and free users will ever see Argon, and Google's announcements set no free usage allowance. Even paying customers are getting it in stages: an Android Police editor subscribed to AI Pro still had no access on launch day. Enterprise early testers include Deloitte, LexisNexis, Ambience Healthcare and Isomorphic Labs. Google also said it would give 10,000 scientists the highest usage tier in Antigravity free for six months, onboarded on a rolling basis.

What Comes Next: Higher Prices, More Models

Three things to watch. First, the bill: after November 30, Argon's per-token cost doubles and developers have to redo their math. Second, reach: after Pro and Ultra, there is no date for Plus, free users or additional regions. Third, the family: Doshi said explicitly that Argon is only the first Gemini 4 model, with more to follow. Google is already using it internally, saying it freed up hundreds of terabytes of memory at its data centers without buying hardware. For rivals, a frontier model priced at roughly a fifth of the competition makes the API price war tighter heading into the end of the year.