Google Launches Gemini 4 Argon Globally — the Introductory Price Doubles After November 30
Google launched its flagship model Gemini 4 Argon globally on October 13, 2026, after months of delays and a limited pre-release to selected security partners. But what may matter most to developers right now is a dated fact: the introductory price of $2 per million input tokens and $10 per million output tokens only holds until November 30. After that, the prices double to $4 and $20, according to Ars Technica.
What Changed
According to Google's own announcement, Argon is now globally available to developers and enterprises via the API in AI Studio and on the Gemini platform, to Google AI Ultra subscribers in the Antigravity coding tool, and to AI Ultra and AI Pro subscribers in the Gemini app and AI Mode in Search. In addition, the model is available to enterprise customers on Google Cloud, where Google is positioning it for long multi-step tasks such as financial reconciliation, legal research, and cybersecurity tools like CodeMender.
The launch differs from what preceded it: according to Business Insider, Argon was first showcased about a month earlier, but then only to a group of "trusted partners" who were to proactively test the model and uncover cyber vulnerabilities. Now the gates are open — but payment is a requirement.
Who Gets Access — and at What Price
For ordinary users, the threshold is a paid subscription tier. Ars Technica reports that Argon in the Gemini app is only rolling out to AI Pro ($20 per month) and AI Ultra (from $99 per month), while Antigravity and AI Studio are reserved for AI Ultra subscribers. Business Insider puts the minimum price at $19.99 per month for Gemini app users, who also encounter Argon in AI Mode within Google Search.
What about Plus subscribers and free users? The Verge reports that it is unclear when — or whether — Argon will become available to these groups. Google has said nothing, and there is so far no promise of expanded access.
The Pricing Clock: $2 and $10 — Until November 30
For developers, the concrete news is the pricing structure. Google's announcement lists an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens at 95 percent off. According to Ars Technica, the offer only holds until November 30; after that, prices rise to $4 and $20 per million tokens, respectively.
For teams building on the model, that means a clear decision gate: anyone who wants to test Argon seriously has a financial incentive to do so before the deadline. Cached tokens at 5 percent of the regular price also make repeated calls with long, stable contexts — typical of agent-based workflows — substantially cheaper during the introductory period.
The Verge characterizes the introductory price as roughly one-fifth the cost of OpenAI's GPT-6 Astra and Anthropic's Fable 5. It is worth noting that this is the outlet's analysis, not an independently verified fact — and it is unclear whether the price after November 30 will still sit below competitors' levels. The comparison also applies only to the introductory price, not the level the model lands on December 1.
The Competitive Backdrop: Delays and Reorganization
Reuters frames the launch as an attempt to regain ground against OpenAI and Anthropic after months of delays and a leadership change at the DeepMind lab. According to Il Sole 24 ORE, Argon was announced in May at Google I/O with a planned June release — and thus arrives several months behind schedule, after both Anthropic and OpenAI released their latest flagship models, Fable 5 and ChatGPT-6.
The sources do not, however, give a fully consistent picture of the timeline before launch. While Il Sole 24 ORE places the announcement in May, other outlets describe the unveiling as recent. These accounts cannot be fully reconciled from the available reporting, but it is well documented that the model arrived late and that competitors got there first. The Verge characterizes Argon as Google's first major frontier model release since Gemini 3 in late 2025.
The Performance Claims — For Now, Only Google's Own Numbers
Several of the most striking claims about Argon rest solely on Google itself:
- Vals Index: Google claims Argon ranks number one on this index, which measures model performance on multi-step tasks in sectors such as finance, coding, law, and tax.
- Harvey's Legal Agent Benchmark: Google states that Argon leads this evaluation for legal research and drafting, with strong source fidelity when teams evaluate complex legal precedent.
- Internal coding effect: Google says that using Argon in Antigravity has increased agentic code submissions by more than 35 percent and reduced code rollbacks by nearly a third. Il Sole 24 ORE adds details from Google: the figures are said to come from the first four weeks of use among over 100,000 Google employees, where 30 percent fewer rollbacks means less need to undo changes and revert to stable versions.
None of these results have been independently verified in the available reporting. They should be read as the company's own marketing figures — relevant as indicators of what Google itself believes the model is good at, but not as established facts.
Safety Measures
Google says it has stepped up efforts to prevent the model from behaving in unwanted ways by monitoring the model's chain-of-thought reasoning and halting processing if an automated monitoring system detects an unintended workflow or derailment. Tulsee Doshi, product lead for Gemini at Google DeepMind, has according to Business Insider characterized Argon as a "powerhouse" model and said it will be the first in a family of new Gemini models to roll out over time.
What Remains Open
Three questions are open after launch. First: will Plus subscribers and free users ever get access? Google has not said. Second: will the doubled price after November 30 still keep Argon cheaper than GPT-6 Astra and Fable 5? No available source answers that. Third: will the benchmark results and productivity figures be verified by independent parties? So far, everything comes from Google itself.
For developers and enterprises, though, the decision is concrete and time-bound: Argon is available globally now, at a price that expires on November 30.