TL;DR: Google started wide rollout of Gemini 4 Argon on Oct. 13, 2026, limiting it to paid Google AI Pro and Ultra subscribers in the Gemini app and to Ultra subscribers for Antigravity coding workflows.

Paid rollout begins after cautious preview

Google revealed Gemini 4 Argon a few weeks earlier and said it needed to move cautiously toward a full release because of the potential for nefarious use, according to Ars Technica.

That caution period ended quickly. Argon began its wide rollout on Oct. 13, 2026, but only for paying users at this time, with even lower-tier paid plans lacking full access in coding-focused products, according to Ars Technica.

The timing matches Google's own announcement. In a post titled "Gemini 4 Argon is now available in the Gemini app," Google said on Oct. 13, 2026, that Gemini 4 Argon is now available in the Gemini app, delivering frontier reasoning and improved performance across complex tasks.

A separate Google announcement titled "Gemini 4 Argon: Built to solve your hardest problems" said Gemini 4 Argon is now globally available today for developers and enterprises via the API in AI Studio and Gemini platform, for Google AI Ultra subscribers in Google Antigravity, and Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search.

That distinction matters for consumers. General chat access is broader than agentic coding access, a split Google maintained across its consumer and developer surfaces on launch day.

Who gets Argon in the Gemini app

For general AI queries, Gemini 4 Argon rolled out in the Gemini app for those on AI Pro at $20 per month and AI Ultra starting at $99 per month, according to Ars Technica.

Google's Gemini app announcement said Google AI Pro and Ultra subscribers can now access Gemini 4 Argon and give Gemini tasks to get more done, with availability starting that day.

Google described Argon in the app as answering questions more directly with less hedging, repeating itself no more than necessary, and skipping cliches throughout responses. For longer tasks, Google said Argon keeps users in the loop by sharing its progress more often and more effectively.

The release is also part of a broader shift in Google's free-versus-paid lineup. Google also recently restricted access to the Flash models for free users, leaving them with just Flash-Lite, according to Ars Technica.

That context explains the headline limit: free Gemini users did not get Argon on Oct. 13, while Pro and Ultra subscribers did in the main chat experience.

Why Antigravity access is narrower

Coding-focused access is more restricted. In Antigravity and AI Studio, the new model will only be available for those on AI Ultra plans, according to Ars Technica.

The reason cited is cost and capability. Code generation has become even more important in frontier AI development since Google's last Pro model release, but it can consume substantial compute cycles, according to Ars Technica.

Google said users can generate code with Argon in the Gemini app's Canvas mode even as a Pro subscriber, but the multi-threaded agentic workflows in Antigravity are likely too expensive to offer as broadly, according to Ars Technica.

Google said it is keeping Gemini 3.1 Pro around for the time being for users who need something more capable than Flash models, according to Ars Technica. Google declined to clarify when, or if, Argon will come to Antigravity for other subscription tiers, according to Ars Technica.

The Antigravity team separately said Gemini 4 Argon was rolling out directly inside Google Antigravity for Google AI Pro and Ultra subscribers in the Antigravity desktop app and Antigravity CLI, with business availability via API and eligible Gemini Enterprise subscriptions. Google's central Oct. 13 model announcement, however, listed Antigravity access for Google AI Ultra subscribers.

What changed in the Gemini app experience

Google framed the Gemini app update as both a model swap and a workflow change, according to its post titled "Gemini 4 Argon is now available in the Gemini app."

Argon brings major improvements in coding, and Google said it is making it easier to switch between chatting and coding with Google Antigravity's agentic coding interface now available directly in the Gemini macOS app.

Google also said it is easier to use Gemini from the main chat window, with a simple toggle from chat to tasks. When users choose tasks, Google said Gemini can work even when the user is not at the computer, helping get more done across Google apps and more.

Google said Gemini now connects to more than 40 apps — like Instacart, OpenTable, Adobe, and Shopify — to help tackle tasks from ordering groceries and booking tables to designing and building a business.

Those app connections position Argon less as a simple answer engine and more as a task-execution layer inside the consumer Gemini experience.

Capabilities, safety controls and benchmarks

Nothing about Gemini 4 Argon's capabilities changed at rollout. Google still pointed to benchmarks it published late last month as evidence of advanced reasoning and agentic coding capabilities, according to Ars Technica.

Google said the model is trained to refuse malicious instructions, and an automated misalignment monitor can stop processing if it detects an unintended workflow, according to Ars Technica. Google claims Argon needs those controls because it is so capable, according to Ars Technica.

In its central launch post, Google said Argon excels at deep work in complex, real-world tasks across domains including coding, finance, legal, cybersecurity and science. Google said its depth and precision across multi-step tasks leads Argon to outperform other frontier models on benchmarks across domains, ranking No. 1 on the Vals Index, which measures model performance across finance, coding, legal and tax work.

Google also highlighted spatial reasoning. Ars Technica reported that Argon is apparently better at solving spatial problems in 3D environments, illustrated by Google with a 3D example image.

The Verge reported on Oct. 13 that Google claimed Argon delivers frontier reasoning and improved performance across complex workflows while cutting down on hedging, repetition and cliches, and that Google claimed Argon outperforms rival models including GPT-6 Astra.

Internal Google results cited at launch

Google offered several internal examples of Argon use, where it said the model had been used by thousands of Googlers, according to Ars Technica.

One example involved YouTube's comment system. Argon reportedly was instrumental in making the system more efficient after Argon identified redundant emoji rendering that, when eliminated, dropped server load by 45 percent, according to Ars Technica.

Engineers also turned Gemini 4 loose on the proprietary drone physics simulator used in Project Wing, and the AI found critical bugs that had been in the codebase since 2018, according to Ars Technica. Google also noted that Argon advanced its robotics program, showing a measurable step increase in embodied reasoning, according to Ars Technica.

Google's Antigravity blog gave a more detailed version of the YouTube work, saying Argon in Antigravity helped cut the CPU footprint of YouTube's comments frontend in half. By tracing production profile data, Google said Argon identified that redundant emoji rendering alone accounted for nearly 45% of server overhead, leading to 12 targeted optimizations.

The same Antigravity account said those updates reduced server comment-load latency by 30 ms and dropped client end-to-end latency by 20 ms.

For Wing, the Antigravity blog said Argon reviewed Wing's proprietary internal drone physics simulator, reasoning through the underlying physics and math rather than just code structure. Google said it highlighted areas for improved physics simulations and proposed 18 targeted modelling improvements across aerodynamics, propulsion and structural dynamics.

Scale of internal testing and enterprise path

Google said its most honest proving ground was its engineering teams before bringing Argon to developers in Antigravity, according to the Google Antigravity Blog.

Google said over 100,000 users across Alphabet rely on Argon inside its internal version of Antigravity each month to solve complex production problems. Over a four-week period, Google reported a 35% increase in agentic code, alongside improved code quality with a 30% decrease in post-submit rollbacks.

Other internal projects cited included a backlog effort in which Argon reviewed 134 open bug reports, closed 97 stale or duplicate bugs with documented commit proof in under three hours, and sorted the remaining 37 by owner. Google described that as a 72.4% queue reduction.

Google also cited legacy modernization work, including mapping and rewriting fragile Fortran code into modern Rust with a 1.8-times performance boost, plus a security audit in which Argon found four critical security bugs in access-control changes after two earlier frontier models found none.

For businesses, Google said Gemini 4 Argon is available via the API in Gemini platform and within Antigravity on eligible Gemini Enterprise subscriptions, according to the Google Antigravity Blog.

Developer API pricing and deadline

For developers accessing Google's models via the paid API, Google offered temporary Argon pricing, according to Ars Technica.

Pricing was temporarily set at $2 per million input tokens and $10 per million output tokens, with a 95 percent discount for cached tokens, according to Ars Technica. Google's central announcement likewise listed an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens at 95% off.

The introductory deal is only good through Nov. 30, when the rate will rise to $4 per 1 million input tokens and $20 per 1 million output tokens, according to Ars Technica.

Developers building workflows around the new model therefore face a near-term pricing decision, with the lower rate expiring about six weeks after launch.

Sources referenced