Google has finally released its next-generation frontier AI model, called Gemini 4 Argon. According to the company, Argon brings in major improvements in various tasks such as coding, professional work, and cybersecurity compared to its previous AI models.
Gemini 4 Argon’s release comes at a time when rumours emerged that Google was falling behind rivals, including OpenAI and Anthropic, in the AI race after reportedly delaying Gemini 3.5 Pro. On X, Google CEO Sundar Pichai addressed such speculation. "Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible,” he wrote. “Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering.”
Sundar Pichai added that teams within Google were already using Gemini 4 Argon for different tasks including coding and quantum computing. It is also the first time Google has used the term Argon for an AI model.
How good is Gemini 4 Argon?
Google describes Gemini 4 Argon as a frontier AI model built for deep reasoning across complex workflows. Argon’s output token limit stands at 1 million tokens, which Google says, allows it to think more deeply and generate hundreds of thousands of tokens in a single trajectory for longer and more complex use cases.
The company claims that Argon can match or outperform some of the most advanced AI models – such as GPT-6 Astra, Claude Fable 5.1 – in certain tasks. In benchmark scores posted by the company, Gemini 4 Argon scores higher in multiple tests. In the knowledge work benchmark Vals, the Google model scores 68.9 per cent, compared to 63.1 per cent for GPT-6 Astra and 65.8 per cent for Fable 5.1
In coding too, Gemini 4 Argon can outperform rivals. On Vibe Code Bench, Argon managed to get a score of 91.9 per cent. On the other hand, Claude Fable 5.1 stood at 90.3 per cent with GPT-6 Astra scoring 89.6 per cent. Though in certain cases Argon was a bit behind other models such as Terminal-bench 4.0.
Gemini 4 Argon’s cybersecurity capabilities
As a new generation frontier model, Gemini 4 Argon comes with advanced cybersecurity capabilities. Google says Argon was trained to be “highly capable at cybersecurity defense” and can autonomously find, validate and patch critical software vulnerabilities.
Wiz, a Google-owned cybersecurity firm, is already using Argon through its Scan for Good initiative. In an early demonstration at Wiz, Argon found a critical vulnerability exposing sensitive personal information across healthcare software used by hospitals worldwide that previous frontier models had missed.
Gemini 4 Argon gets limited release for now
Argon’s release comes at a time amid growing talk about AI safety. Google says it is releasing the model to a set of trusted cyber defenders through Google’s Fairwind programme and the US government for now. Similar to OpenAI’s Daybreak or Anthropic’s Project Glasswing.
"Importantly Argon has frontier safeguards and we are rolling it out responsibly - it’s with the US govt and going to a set of trusted cyber defenders through our Fairwind Program today,” Sundar Pichai wrote on X. “We’re going to make it available as soon as we can and as safely as we can. So hold tight, lots more coming, and you’re going to see us iterating rapidly." Google is one of the big tech firms that signed US President Donald Trump's AI accords.
Broader access for Argon will begin with paid API customers and Google AI Ultra subscribers, after feedback from trusted testers and cyber defenders. Google says Argon will launch at an introductory price of $2 (roughly Rs 190) per million input tokens and $10 (roughly Rs 950) per million output tokens, with cached input tokens priced at a 95 per cent discount. After the introductory period, the price will be $4 (roughly Rs 380) per million input tokens and $20 (roughly Rs 1900) per million output tokens.
