Redazione Zero Sections IT ES EN

Updated at 21:01 (Italian time) 7 Nov 2026

Tech & AI · Analysis Wednesday, 14 October 2026 · Morning edition, 6:30 · AI-generated content, without human review

Google opens Gemini 4 Argon to businesses and developers

After a phase reserved for cybersecurity defenders, Google's new model becomes accessible via API, AI Studio and Antigravity, with an introductory price that will double at the end of November.

Fotografia d'archivio, non riferita ai fatti descritti nell'articolo
Immagine d'archivio, non riferita ai fatti descritti. Foto di panumas nikhomkhai su Pexels

Google made Gemini 4 Argon available to professional users starting October 13, 2026. The model, until then reserved for a restricted group of cybersecurity specialists, is now accessible within the Gemini app for Pro and Ultra plan subscribers, in the paid API, and in the Google AI Studio and Antigravity development environments, the latter two limited to Ultra plan users.

The choice to open first to a circle of cybersecurity defenders and then to the broader market follows a logic Google has already used for other releases: testing a model on users with high technical skills, capable of identifying flaws or anomalous behavior before the system reaches a wider audience of developers and businesses. No details have been disclosed about how long this reserved phase lasted or what emerged during use by the specialists.

The most concrete part of the news concerns pricing. During the introductory period, API usage costs $2 per million input tokens and $10 per million output tokens. This is a temporary rate: starting November 30, 2026, the cost will rise to $4 for input tokens and $20 for output tokens, a clean doubling compared to the launch price list. The two-tier structure — cheaper at the start, then aligned to a “standard” price — is a common practice among large language model providers, designed to encourage initial adoption by developers and businesses building products on top of the API, before stabilizing margins once integration has taken place.

For businesses considering building applications on Gemini 4 Argon, the window until November 30 therefore represents a reduced-cost period in which to experiment at scale before the bill grows. The distinction between the price of input tokens and output tokens — the latter increasingly more expensive, here five times as much — reflects the greater computational cost of generating text compared to simply reading it on the model’s part, a ratio found in many industry price lists.

It remains to be seen how Gemini 4 Argon compares to competing models in terms of performance, since the material currently available concerns exclusively commercial availability and pricing, not benchmarks or technical comparisons. Availability on Antigravity, a development environment less known to the general public compared to AI Studio, suggests particular attention from Google toward the more advanced, paying users of the Ultra plan, who are given the broadest access to the tools.

It should be noted that the news currently comes from a single source — an article based on a Google press release — and there are currently no independent confirmations of the pricing and availability details reported here. This is commercial and technical information, verifiable in the coming days through actual use of the product by third-party developers, who could confirm or correct the figures released by Google itself.

The dual pricing, introductory and standard, remains for now the most verifiable and concrete element of the entire operation: $2 and $10 per million tokens until November 30, 2026, then $4 and $20 on a stable basis.

← Archive · Front page · Past editorials · Report an error · Original article (in Italian)