Gemini 4 Argon reaches paying users: what’s available, and where

The change: a global rollout, but not for everyone

On October 13, 2026, Google began rolling out Gemini 4 Argon in the Gemini app for Google AI Pro and Google AI Ultra subscribers. The official release note describes a global rollout for those two plans. It does not say that every Gemini user will receive the model, or that every feature will be available in every country at the same time.

That distinction matters: calling a rollout global does not mean access is universal. Google’s plan page for Spain includes Argon in Pro and Ultra, but assigns different levels of access and warns that some benefits depend on the country. Before assuming a feature is active, check the specific product, account, plan and region.

Where to find Argon, and how access differs by channel

The Gemini app has the clearest confirmation of a start date: the official release notes say the global rollout began for AI Pro and AI Ultra subscribers. Google’s Spain plan page also lists Argon in both tiers, with expanded access for Pro and greater access for Ultra. Those descriptions indicate different levels of access; they do not specify a particular number of queries or guarantee identical availability for every account.

Google also lists Argon in its plan information for Google AI Studio and Google Antigravity, under different terms. The AI Studio table compares model access limits; the Antigravity table compares usage limits. The documentation available here does not establish one shared activation date for every channel, or confirm that all free developer accounts can select Argon. For Search, the Spain plan page describes expanded access to the model in AI Mode for Pro subscribers and a higher level of access for Ultra. That does not mean unlimited availability either.

Plans, accounts and restrictions: who can get access

For individual users in Spain, Google’s official information lists Argon in Google AI Pro and Google AI Ultra. AI Pro includes access to the model for complex workflows; AI Ultra provides a higher level of access. Google’s plan page says Argon is available to subscribers of those plans in certain countries. That does not prove it is enabled for every eligible account at the same time: a gradual rollout may take time to appear.

The same page says Google AI plans are intended for personal accounts. Google Workspace customers should check the add-on options for their business subscription, rather than assume that a work account has the same rights as a personal Pro or Ultra account. Age requirements, system requirements, language restrictions and regional differences may also apply. The fact that a plan is available internationally does not, by itself, prove that every included feature is available in every market.

Benchmarks: useful signals, not conclusive proof of superiority

The initial Argon announcement included scores from software engineering, cybersecurity and professional-task evaluations. But results from different tests cannot be ranked as though they measured a single capability. Tasks, datasets, scoring criteria and test conditions vary. Even two coding benchmarks may assess different skills, such as fixing issues, writing code or completing long-running workflows.

In addition, some scores shared in connection with the launch come from evaluations selected or described by Google. Without general access to the model and reproducible external methodologies, those results cannot establish an independent conclusion about everyday performance. The evidence consulted here confirms plans, channels and the rollout date, but does not include enough independent reports to verify benchmark comparisons. Claims that Argon leads or outperforms competing models should therefore remain attributed and provisional.

Security and rollout: what is documented, and what remains to be checked

Google DeepMind has published information about Argon’s responsible and safe development. It is relevant context for understanding the security approach and the initial restricted rollout. That material does not, on its own, establish the commercial scope of the later rollout on October 13; Gemini’s release notes and Google’s plan page are more directly relevant for that. Keeping these sources of evidence separate avoids confusing a controlled testing phase with later availability in consumer products.

The information available does not establish that every safeguard described for the initial phase remains unchanged across all products, or specify how safeguards vary by region, account or task. Nor does it independently verify every cybersecurity capability attributed to the model. For businesses, commercial availability is not a substitute for a risk assessment: before connecting Argon to code, sensitive data or production workflows, verify permissions, activity logging, human review and the data-handling terms that apply to the chosen channel.

What to check before trying it

For individual users, the practical steps are to confirm that the subscription is Pro or Ultra, check whether Argon appears in the model selector, and verify that the specific feature is available in their country. Developers should check AI Studio and Antigravity separately: a plan that includes expanded access to a platform does not guarantee that every model or tool has the same limits.

Organizations should base decisions on their own tests, not just a published score. Start with non-critical tasks, measure cost and reliability, restrict system permissions, and require human review before making significant changes. If the organization uses a business account, also confirm which service terms and data protections apply. Argon’s availability is a meaningful change, but its practical value will depend on the specific product conditions and the results of evaluation in each use case.

Keep exploring

Corrections and transparency

If you spot incorrect or outdated information, send us a correction with the page and source we should review.