Google’s Gemini 4 Argon launches cheap, then doubles in price Nov. 30

Google’s newest AI model comes with a sale sign and an expiration date. Developers can use Gemini 4 Argon at an introductory price until Nov. 30. After that, the price doubles.

Google released Gemini 4 Argon to the public following a two-week preview period, at an introductory price of $2 per million input tokens and $10 per million output tokens, Forbes reported. Those prices are set to double after Nov. 30, according to the same report. Google’s Gemini release notes list an Oct. 13 entry introducing Argon as its frontier model, rolling out globally to Google AI Pro and Ultra subscribers in the Gemini app.

Cyber partners first

The rollout came in stages. When Google first released Gemini 4 Argon on Sept. 30, it went only to a select group of cybersecurity partners through a program Google calls Fairwind, TechCrunch reported. TechCrunch described it as a model aimed at coding, research and writing, which Google called its most powerful yet.

That sequencing says something. Handing a new frontier model to security partners before the public suggests Google wanted defenders to test it, and perhaps to get a head start, before attackers could use it too. It mirrors a broader pattern this month: OpenAI rated its newest GPT-6 models “High” for cybersecurity capability before releasing them, as AIME 360 reported.

“As soon as possible”

The release came faster than Google had suggested even weeks earlier. In late September, DeepMind’s Koray Kavukcuoglu said Gemini 4 was in post-training, the final stage of refining a model before release, and that it would arrive “much earlier” than the end of 2026, according to 9to5Google, which reported Google saying the release was coming “as soon as possible.”

That urgency is the story of the AI race in 2026. Anthropic, OpenAI and Google have each shipped major models within weeks of one another, and every company is under pressure not to fall behind for long.

The price is the pitch

Forbes framed the launch around price, and for good reason. An introductory rate that doubles on a fixed date is a classic tactic: get developers to build on your platform while it’s cheap, and count on them staying when it gets more expensive because switching takes work.

For a startup choosing which model to build on, that’s a real dilemma. A cheap rate for six weeks doesn’t matter much if the long-term cost is higher than a competitor’s. Engineers comparing Argon with rivals have to look past the launch price to what they’ll pay in December.

It also shows how quickly AI pricing is moving. Earlier this month, Anthropic cut the price of its smallest model by 90%, and VentureBeat compared that price with OpenAI’s GPT-6 Luna. Google, with Argon, is competing at the top end, where models cost more but promise more.

There’s a strategic layer, too. Google controls a search engine, a mobile operating system, a cloud business and an office suite. Every developer who builds on Gemini is also a likely customer for the rest of that stack. A discounted model can pay for itself several times over if it pulls a company deeper into Google Cloud.

From the subscriber’s chair

For ordinary users, Argon arrives through Google AI Pro and Ultra subscriptions in the Gemini app, according to the release notes. That means the most capable version sits behind a paywall, while free users get older or smaller models. That split, free basic AI and paid frontier AI, is becoming standard across the industry.

Most people won’t notice the token prices at all. They’ll notice whether the answers get better, faster or more reliable, and whether the monthly subscription feels worth it.

What we can’t verify

Benchmark claims around the launch vary widely in quality. At least one site published a benchmark comparison that AIME 360 couldn’t verify, and an explainer page about Argon’s availability predated the Oct. 13 rollout, so it’s already out of date. Company performance claims, from Google or anyone else, should be treated as claims until independent testers check them.

Google also hasn’t published, in the coverage reviewed for this story, a single launch announcement laying out every detail. Much of what’s known comes from release notes, partner programs and press reports.

The deadline that matters

Nov. 30 is the date to watch. If developers flock to Argon before the price doubles and stay afterward, Google’s strategy works. If they test it cheaply and drift back to rivals, the introductory rate will have been an expensive experiment.

Either way, the AI race now runs on two clocks: how fast companies can ship the next model, and how long they can afford to sell it at a discount.

Sources

• Forbes: Gemini 4 Argon launch puts a deadline on its price advantage
• Google: Gemini Apps release notes
• TechCrunch: Google releases Gemini 4 Argon
• 9to5Google: Gemini 4 release coming “as soon as possible”
• VentureBeat: Claude Haiku 5.5 price compared with GPT-6 Luna

Photos

• “Abstract Blue Glowing Lines on Dark Background” by amynapaloha, Envato Elements (featured image)