Gemini 4 Argon: paid access, pricing, and the arrival of Carbon

  • Gemini 4 Argon was opened to the public on October 13 for Google AI Pro, Ultra and the paid API.
  • The initial price is $2 per million tokens entering and $10 per million exiting, with a subsequent increase to $4 and $20.
  • Google Cloud is geared towards long-term business workflows and multi-stage tasks.
  • Gemini 4 Carbon, announced on October 28, scores 59 points on the v5.0 index but remains limited to cyber defense.

Gemini 4 Argon artificial intelligence model

Google has stopped treating Gemini 4 Argon as a lab promise and has placed it at the heart of its paid offerings. The model was opened to the public on October 13, 2026, after an initial phase reserved for cybersecurity specialists, and has since appeared in the Gemini app, Search's AI mode, and the API for paying developers.

The company is thus maintaining a two-pronged approach: Argon is the Gemini 4 that is already available for use, while Gemini 4 Carbon is presented as a more powerful, later step, although for the moment it is limited to cyber defense equipment. This coexistence of an available model and another announced with restricted access explains much of the buzz surrounding the Argon name.

Gemini 4 Argon
Related article:
Gemini 4 Argon: Google deploys its advanced AI for professionals

A release that's already on the street

Argon was initially announced on September 30, 2026, and its commercial rollout began on October 13. Google AI Pro and Google AI Ultra are the plans that enable its use in the Gemini app and Search's AI mode; the paid API and Google Cloud Enterprise solutions complete the entry point. The sources consulted do not place Argon in the free plan or in Google AI Plus.

Prices: the entrance fee is not free

For developers, Argon launches with an introductory fee of $2 per million tokens purchased and $10 per million tokens sold . After this initial period, the price will increase to $4 per million tokens purchased and $20 per million tokens sold. The cost of caching is reduced by 95%, and Google has indicated that Carbon will launch with the same pricing structure as Argon, making it clear that access to the newer model is not intended to be free.

Gemini 4 Argon
Related article:
Gemini 4 Argon: Google limits its most powerful AI to cybersecurity

Long tasks and business work

Argon's approach isn't so much about quickly answering a single question as it is about sustaining long-term, multi-stage processes . Google describes it as a model for deep reasoning in complex, long-running workflows, and Google Cloud focuses on automating business processes that previously required constant monitoring. The idea is for the system to maintain context, apply specialized knowledge, and reduce execution times for tasks that can take hours or days.

In the business world, this approach aligns with processes in human resources, customer service, document analysis, and software development. The model isn't presented as a magic bullet : Google emphasizes that it can report when a task isn't completed or when the result is unfavorable, and that it's designed to respect the limitations of the systems it works with. For many organizations, this combination of autonomy and control is more valuable than a one-off improvement in an isolated test.

Cybersecurity and controlled deployment

Argon wasn't launched open to everyone. Before October 13th, it underwent rigorous testing by a group of verified cybersecurity specialists , a process Google has repeated with Carbon. In the official narrative, security is presented as a cornerstone of the model: Google Cloud highlights its performance against indirect instruction injection attacks and integrates it into protection tools like CodeMender and critical infrastructure analysis initiatives. This is a cautious approach, because the model remains a paid technology with conditional access.

Close-up of a humanoid robot with a digital face, representing modern artificial intelligence and the Gemini model ecosystem.
Related article:
Gemini Flash vs Flash-Lite: A Complete Guide to Differences and Uses

Gemini 4 Carbon, the next step

On October 28, 2026, Google announced Gemini 4 Carbon, the second model in the Gemini 4 family. Carbon scored 59 points on the Artificial Analysis Intelligence Index v5.0 , a figure the company presents as a benchmark of its progress, but its availability remains very limited: for now, it is reserved for cyber defense teams in the Fairwind program and has no public release date. For the average user, Argon continues to be the true entry point to Gemini 4.

What does this mean for Spain and Europe?

The sources consulted do not include a specific timeline for Spain, but the rollout of Argon is planned globally for Pro and Ultra subscribers , as well as those using the paid API. This means that a Spanish user with Google AI Pro or Ultra can find it in the Gemini app and in Search's AI mode, while the free plan and Google AI Plus are excluded from this initial rollout. In Europe, the key factor is the subscription type and enterprise access, not a different date for each country.

The underlying takeaway is that Google is increasingly separating its edge model from the rest of its product range. Argon is reserved for those who pay for or subscribe to cloud services , Carbon is first tested in sensitive environments, and the Flash and Lite models cater to the casual user. This stratification aligns with a market where the price per token and the ability to automate lengthy tasks are as important as scores on a benchmark leaderboard.

Based on what's been confirmed, Gemini 4 Argon is already functioning as the benchmark paid model within Gemini 4, with launch prices that subsequently increase, enterprise use on Google Cloud, and a global rollout for Pro and Ultra. Carbon doesn't change that picture for now : it adds a new name to the catalog and raises the figure in the Artificial Intelligence Analysis index, but it remains in the hands of cybersecurity and has no public release date. For Spain and Europe, the practical question isn't when Argon will arrive, but whether the contracted plan or the paid API justifies its cost compared to Google's own alternatives.

Gemini 4 Argon
Related article:
Gemini 4 Argon: access, price and tests it has passed

Add as preferred source in Google