Google has rolled out Gemini 4 Argon, its newest frontier AI model, to users worldwide. “Gemini 4 Argon is here,” Google CEO Sundar Pichai wrote in a post on X, adding, “Hope you all enjoy it!”
Availability
Developers and enterprises can access Argon through the paid tier of the Gemini API, AI Studio and the Gemini platform. Google AI Ultra subscribers can use it in Google Antigravity, while both Ultra and Pro subscribers get it in the Gemini app and AI Mode in Search.
For developers, Argon is priced at an introductory rate of $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95 per cent. Through the Gemini API, it supports a combined input and output limit of 1 million tokens.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
ALSO READ | Gemini 4 Argon: Google Unveils New Frontier AI Model — Prices, Features And Key Things Explained
Built for Complex Tasks
According to Koray Kavukcuoglu, SVP at Google DeepMind and Google's Chief AI Architect, Argon excels at demanding, multi-step work in areas such as coding, finance, law, cybersecurity and science. Google says the model holds the top spot on the Vals Index, which tests AI performance on economically important work such as legal, tax, coding and finance tasks.
Inside Google, engineers have already been putting Argon to work on difficult technical challenges. Teams using it in Antigravity submitted over 35 per cent more AI-generated code, while the number of changes that had to be reversed fell by almost a third. It also helped the Chrome team complete a major code restructuring across hundreds of files in weeks instead of months.
Science, Robotics and Cybersecurity
Google says Argon has supported research in quantum computing, nuclear fusion, rare disease diagnosis and enzyme design. The company is offering 10,000 scientists free access to Antigravity's highest usage tier for six months, subject to eligibility, capacity and selection, and applications are now open. In robotics, Argon scored 87 per cent on the ERQA embodied reasoning benchmark, ahead of other frontier models, according to Google.
On the cybersecurity front, Google is expanding access to hundreds of additional partners in its Fairwind Program, which gives vetted defenders access to Argon without cyber guardrails.
ALSO READ | Google Gemini Live May Add 'Ephemeral Mode' For Unsaved Video Interactions: Reports
Safety Measures
Google describes Argon's safety setup as its most comprehensive yet. When outside experts tried to push the model into helping with risky biology tasks, they succeeded on fewer than 1 per cent of the tests. On coding benchmarks, the model attempted unauthorised shortcuts just 2 per cent of the time. Argon also leads the Gray Swan prompt injection benchmark, with attacks succeeding just 0.7 per cent of the time across 15 adaptive attempts.
Google says more models in the Gemini 4 family will follow.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
