Google Gemini 4 Argon Now Available Worldwide: Price, Features

Google has opened up Gemini 4 Argon, its newest and most capable AI model, to developers and businesses around the world. The worldwide rollout began on October 13, two weeks after the company first unveiled the model and gave early access only to a small group of trusted cybersecurity defenders. For Indian developers, startups and paying Gemini users, it is the first chance to try Google's new top-tier model for themselves.

Argon is the first model to carry the Gemini 4 name. Google is pitching it as a model built for long, complicated jobs rather than quick chatbot replies, including writing and fixing large codebases, financial and legal analysis, and security work.

From a limited preview to a global launch

Google DeepMind's Koray Kavukcuoglu introduced Argon in a company blog post on September 30, calling it Google's new frontier model. At that stage it was going only to cyber defenders through what Google calls the Fairwind Program, with a promise of a wider release starting with paid API customers and Google AI Ultra subscribers.

That wider release is now under way. According to reports on the launch, paying customers can reach Argon through the Gemini API in Google AI Studio and through Antigravity, Google's agent-based coding platform. Subscribers to the Google AI Pro and Ultra plans can use it inside the Gemini app.

What Argon is built to do

Google says Argon reaches top-level performance on real-world software engineering, enterprise knowledge work in law and finance, and cyber defence. It is also meant to be strong at visual tasks such as reading charts, understanding long videos and working through stacks of documents. The company even describes it as a partner for creative writing.

The benchmark numbers

Google's own figures put Argon at 77.9% on the DeepSWE v1.1 coding benchmark and 91.7% on LVBench, a long-video understanding test, both of which it calls new state-of-the-art results. On Zapier's AutomationBench, which measures how well a model can run multi-step business workflows, Argon ranked first with 51.3%. These are company-reported numbers, and independent testers will want to check them over the coming weeks.

A much bigger output limit

One change that developers will notice right away is the output limit. Argon can produce up to one million tokens in a single response, compared with 64,000 for earlier Gemini models. In practice, that means it can write or rewrite very large files, long reports or entire code modules in one go rather than in chunks.

How Google says it uses Argon internally

Google shared examples of Argon at work inside the company. Its engineers have used it to help move old C and C++ code to the Rust programming language, including the libgav1 video decoder, where the Rust version ran 2.7 times faster while producing identical video output. Google also says Argon-powered agents found memory savings across its systems that could free up hundreds of terabytes once rolled out.

Pricing: cheaper for now

For the introductory period, API pricing is $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached input. Google's blog says the rate will then rise to $4 and $20 respectively. Reports on the launch say the higher rates begin in December, which gives developers roughly six weeks of cheaper testing.

There is no free tier for Argon at launch. Technology site Gizmodo noted that this may slow down everyday adoption, even though the model is priced well below some rival flagship models.

Safety and security measures

Because Argon is so capable at security work, Google says it has trained it to refuse requests for cyber attacks and for chemical, biological, radiological and nuclear misuse, while still helping with legitimate dual-use research. The company calls it its most resistant model yet to indirect prompt injection, the trick of hiding malicious instructions inside web pages or documents that an AI reads. A version without cyber guardrails is being given only to vetted defenders and Google's internal teams. Google also says it took part in the US government's voluntary pre-release testing process.

What it means for users in India

India has one of the world's largest developer communities, and Google has not announced any separate timeline or restrictions for the country. Indian developers building on the Gemini API can start testing Argon through AI Studio right away, and the introductory price makes this a good window to compare it with the models they already use.

For companies in IT services, fintech and legal tech, the pitch is clear: Google wants Argon to handle long, multi-step tasks such as code migration or document review. Whether it delivers on that will depend on real-world testing, not launch benchmarks. Ordinary Gemini app users on free plans will not see Argon yet.

What happens next

Argon is only the first member of the Gemini 4 family, and Google has not yet said when smaller or cheaper Gemini 4 models will follow. The bigger questions for the next few weeks are how Argon performs in independent tests, how quickly businesses adopt it before the price goes up, and how rivals respond in what has become a fast-moving race between AI labs.

Also Read

• TRAI Report: India Internet Users Top 1.11 Billion, Data Use 28 GB
• RDI Fund Picks 19 Fund Managers for Rs 7,131 Crore Deep-Tech Push
• Musk vs Ambani: Starlink Row Erupts, Centre Says No Monopoly