Gemini 4 Argon Goes Global at $2/$10, Doubling After Nov 30

Google said on 13 October 2026 that Gemini 4 Argon is now globally available. Developers and enterprises get it through the Gemini API, AI Studio and the Gemini platform; Google AI Ultra subscribers get it in Google Antigravity; Ultra and Pro subscribers get it in the Gemini app and AI Mode in Search.

Google says the model launches at an introductory price of $2 per million input tokens and $10 per million output tokens. A footnote to its post says that after the introductory period expires on 30 November 2026, $4 per million input tokens and $20 per million output tokens apply.

What Google Released

Thirteen days earlier, on 30 September, Google had opened Argon to cyber defenders in its Fairwind Program, as we reported. Logan Kilpatrick wrote on X that the model went to Fairwind defenders and early access customers over the last two weeks.

Google says Argon ranks #1 on the Vals Index, which it says measures model performance across sectors such as finance, coding, legal and tax work. The benchmark figures are Google’s own.

In the Gemini app, Google says, a toggle now switches from chat to tasks, so Gemini can keep working when the user is away, and the app now connects to more than 40 apps, including Instacart, OpenTable, Adobe and Shopify.

What Was Posted on X

Logan Kilpatrick of Google posted on 13 October 2026 that Gemini 4 Argon is rolling out globally in the Gemini API, to Pro and Ultra in the Gemini app, and to Ultra subscribers in Antigravity. He writes that the model went to cyber defenders in the Fairwind program and to early access customers over the last two weeks, and mentions “1M token output when the model needs to work more”. Google’s own post describes this as a 1M unified input-output token limit.

Koray Kavukcuoglu, who wrote Google’s launch post, posted that Argon is designed for the hardest coding and knowledge work tasks, and listed where it is available: the paid tier of the Gemini API and the Gemini platform, Antigravity for Ultra subscribers, and the Gemini app and Search for Ultra and Pro subscribers.

Aaron Levie, Box’s CEO, posted that Box tested Argon on its Box AI Complex Work eval for the past week and that it scored 14 percentage points above Gemini 3.1 Pro across all its tests, with examples by industry. This is Box’s own evaluation and his own view, not Google’s figures. His post quotes Sundar Pichai’s launch post.

Ethan Mollick, a Wharton professor who had access to Gemini 4 before launch, posted that it is a big jump over earlier Gemini models and “the first non-OpenAI/Anthropic model to pass the Fable/Mythos/Astra bar”. He writes that it is “quite creative and a good writer” and at least at par with OpenAI and Anthropic in some areas, but lags in others, “including at complex multi-agent work”, and has a tendency to be “too certain about its conclusions and a little lazy”. This is his own assessment, not Google’s.

The Price And Its End Date

Google says cached input tokens are 95% off during the introductory price. Its footnote says the later $4/$20 price applies to the Argon model intended for public access only.

When we covered the introductory price on 3 October, Google’s page did not say how long it would last. The footnote on the 13 October post gives the date: the introductory period expires on 30 November 2026.

By our arithmetic, that leaves seven weeks at half the later rate, from 13 October to 30 November.

What Google’s Infrastructure Chief Says About Serving Capacity

Amin Vahdat, Google’s chief technologist for AI infrastructure, told Sonya Huang on Sequoia Capital’s Training Data podcast (episode published 6 October 2026, from about 13:04) that the target is token-generation capability from a serving perspective, and that “what I’m saying here is not necessarily you have to double the number of flops every six months.” His words: “You have to double the capability of that hardware to generate tokens every six months. And as much or more of that is going to come from software as it is from hardware.” Later in the same interview (from about 52:16) he said Google’s “seven and eight year old TPUs are still at a hundred percent utilization.” These are his remarks from an interview published a week before this launch; he does not discuss Argon or its price in these excerpts.

What Google Says Argon Does

Google says Argon in Antigravity has helped its teams increase agentic code submissions by more than 35% while reducing code rollbacks by nearly a third.

Google says Argon helped the Chrome team separate core webpage handling from a legacy folder of about 6 million lines of browser features inside Chromium’s 40 million-line codebase. It says Argon also cut in-memory session prep time in its Agent Development Kit by 40% in long sessions.

Google says the Gemini API now has a 1M unified input-output token limit, with a new continuation protocol for long tasks. Logan Kilpatrick, in his X post, describes this as 1M token output when the model needs to work more.

Google also says it is offering 10,000 scientists the highest usage tier in Google Antigravity free for six months, subject to eligibility, capacity and selection, and is expanding Fairwind to hundreds of additional partners.

What Box Says It Measured

Aaron Levie, Box’s CEO, wrote on X that Box had tested Argon on its Box AI Complex Work eval for the past week, and that it scored 14 percentage points higher than Gemini 3.1 Pro across all its tests.

He lists gains by industry: technology +37 points, consumer products +33, healthcare +25, financial services +18 and energy +16. These are Box’s own evaluation and his own description, not Google’s figures.

The Structural Read

As we read it, the price is the launch: a frontier model sold at half its own later rate for a fixed window is an offer to move workloads now and to measure the cost later.

As we read it, Google’s own numbers point at long agent work: a 1M unified input-output token limit, a continuation protocol and a tasks mode in the app all allow more tokens per job.

As we read it, Box’s eval is an outside customer’s figure, and it is still that customer’s own test: Aaron Levie reports 14 points over Gemini 3.1 Pro on Box’s Complex Work eval.

Three Implications

DEVELOPERS Google says the introductory price is $2/$10 per million tokens, with cached input 95% off, until 30 November 2026.

ENTERPRISES Google says Argon is available in Gemini for business, the API and Google Antigravity.

SUBSCRIBERS Google says Google AI Pro and Ultra subscribers get Argon in the Gemini app and AI Mode in Search, plus a tasks toggle and more than 40 connected apps.

The Business Engineer Lens

This story maps onto the Business Engineer framework The CFO’s Guide to the Token Economy.

The framework puts it this way: “lower cost per token multiplied by exploding token volume increases total spend”.

As we read it, Argon’s launch fits that pattern: Google cuts the price per token for seven weeks and, in the same release, sets a 1M unified input-output token limit for long tasks, so the bill a customer sees depends on volume as much as on the rate.

What Is Not Established

Google’s posts give no usage, revenue or customer-count figures for Argon. Access for free-tier Gemini app users is not stated in the posts. The benchmark results are reported by Google, and the Box results by Box.

The Bottom Line

Google says Gemini 4 Argon is now globally available across its API, Antigravity, the Gemini app and AI Mode in Search, at $2/$10 per million tokens until 30 November 2026 and $4/$20 after that, per its footnote.