Google Opens Gemini 4 Argon to Developers and Consumers in Wider Release

Google is putting its newest frontier model into products people already use — now it must prove its launch price and benchmark claims in real work

October 14, 2026 — Google has opened Gemini 4 Argon, its newest frontier artificial intelligence model, to developers, businesses and paying consumers in a wider release announced on October 13, 2026. The model is now available through the Gemini API, enterprise services, the Gemini app and AI Mode in Search, competing directly with OpenAI and Anthropic.

Image: A glowing AI model core accessed by developers, businesses and phone users through cloud, search and apps. Credit: AI-generated illustration / Red Eye Eagle News, original illustration.

Editor's Highlight: Argon was first announced September 30 for a restricted group. October 13 changes who can use it: paid developers, enterprise customers and Google AI Pro and Ultra subscribers, at half standard API rates until November 30.

Table of Contents

• Key Facts
• A Wider Release After a Restricted Start
• What Argon Is Built to Do
• Benchmarks Show Strengths and Gaps
• Pricing and the Real Cost Question
• Who Can Access Argon, and How
• Safety, Limits and What to Watch
• Bottom Line
• FAQ
• Sources

Key Facts

A Wider Release After a Restricted Start

Google first announced Gemini 4 Argon on September 30, 2026, with access limited, including to selected cybersecurity professionals.

The October 13 step is about availability, not a first reveal.

Newsroom Focus: Headlines calling Argon simply "new" can mislead — the news this week is that far more customers can now test it in real workflows.

What Argon Is Built to Do

According to Google DeepMind, Argon is designed to maintain reasoning across extended workflows, not only answer single questions — for example investigating a software problem, testing a fix and checking the result.

The model supports text, images, audio and video within a combined window of up to one million tokens (1,048,576 in developer documentation).

Google says Argon uses a sparse mixture-of-experts design, trained on its Tensor Processing Units — only parts of the system run for each request.

A large window helps with long documents and code, but does not guarantee every detail will be interpreted correctly.

Benchmarks Show Strengths and Gaps

Google's published comparisons place Argon ahead on some tests and behind on others. The figures come from Google's model card, so they are results under specific conditions, not a universal ranking.

On the Vals Index for professional work, Argon scored 68.9%, versus 63.1% for GPT-6 Astra, 65.8% for Claude Fable 5.1 and 67.0% for Claude Opus 5.5. On DeepSWE v1.1 it scored 77.9%. On other tests, the order reversed.

Analysis Note: Independent evaluator Artificial Analysis scored Argon at 53 on its Intelligence Index. Settings and time budgets differ across evaluations, so businesses should test Argon on their own tasks.

Pricing and the Real Cost Question

Google is offering Argon at $2 per million input tokens and $10 per million output tokens through November 30, 2026. Google says standard pricing then rises to $4 and $20. Cached input gets a 95% discount during the introductory period, according to the company.

Practical Takeaway: A lower token price does not mean a cheaper finished job. Agents making many calls, using more output tokens, or needing review can cost more in practice.

Who Can Access Argon, and How

Access depends on the product. It is not unrestricted free access.

• Developers: Paid access through the Gemini API and Google AI Studio.
• Businesses: Access through enterprise and Gemini for business services, on contractual terms.
• Consumers: Google AI Pro and Ultra subscribers through the Gemini app and AI Mode in Search.
• Antigravity: The development platform route is for Google AI Ultra subscribers in the consumer rollout.
• Fairwind Program: Approved cybersecurity defenders, first offered in September.

Subscription and API billing are separate: having one does not automatically include the other.

Safety, Limits and What to Watch

Google's model information, as reported in coverage, acknowledges Argon can still produce incorrect information, and identifies risks involving cybersecurity, biological information and oversight, alongside safeguards.

Systems that carry out multiple steps need careful permissions, monitoring and verification, especially with sensitive data. The practical test is whether Argon completes specific tasks reliably, at acceptable cost, after the discount ends.

Bottom Line

Gemini 4 Argon is now an option for far more users than in September, in the leading group on Google's published results, with strengths and gaps.

What to watch: independent testing on real workflows, total cost after December 1, and whether distribution through Search, the Gemini app and Cloud turns trials into sustained use against OpenAI and Anthropic.

FAQ

When did Gemini 4 Argon become widely available?

Wider release was announced October 13, 2026, after a limited September 30 announcement.

Is Gemini 4 Argon free to use?

No. Developers need paid API access, and consumers need Google AI Pro or Ultra, with Antigravity limited to Ultra.

How much does the API cost?

Introductory pricing is $2 per million input tokens and $10 per million output tokens through November 30, 2026. Google says standard pricing then becomes $4 and $20 respectively.

Does Argon beat GPT-6 and Claude on every test?

No. It leads on the Vals Index and DeepSWE v1.1 in Google's comparisons, but trails on FrontierSWE v2 and Terminal-Bench 4.0.

What is Argon designed for?

Google positions it for multi-step professional work — software engineering, finance, legal research, cybersecurity and science — across text, image, audio and video.

Sources

• AI Balkans — Google Releases Gemini 4 to Challenge AI Rivals, wider release, pricing and benchmarks
• INDmoney — Gemini 4 Argon vs OpenAI and Anthropic, access routes, pricing and independent scores
• TechStartups — Top Tech News October 14, 2026, Gemini 4 Argon wider availability summary
• YouTube — World of AI, Gemini 4 Argon release and benchmark testing video
• Image: AI-generated illustration / Red Eye Eagle News, original illustration.