Google opens Gemini 4 Argon globally with US$2 API launch pricing

Google said on 13 October 2026 that Gemini 4 Argon was available globally through the Gemini API and to Google AI subscribers in Google Antigravity, the Gemini app and AI Mode in Search. The Gemini app post names Pro and Ultra plans; Google Antigravity access is limited to Ultra subscribers.[8][1]

Google introduced Argon on 30 September for trusted cyber defenders. Its October announcement describes a wider global rollout for developers, enterprises and eligible consumers; Fairwind remains a separate route for defenders.[3][8][2]

Google says Argon answers questions more directly with less hedging, avoids repetition and skips clichés. On longer tasks it shares progress more often. These are Google's descriptions of model behaviour, not independent test results; the app post gives no controlled comparison or quantified improvement for those features.[1]

The app update also puts Google Antigravity's coding interface inside the Gemini macOS app and adds a toggle from chat to tasks. In tasks mode, Gemini can work across Google apps while a user is away from the computer. Google says Gemini now connects to more than 40 apps, including Instacart, OpenTable, Adobe and Shopify.[1]

Google calls the launch global, but access still depends on the subscription or API route. The posts give no Malaysian ringgit price for Pro or Ultra plans. Users should check their own model picker and account terms before promising the feature to staff or customers.[1][8]

Developers can build with Argon through the paid tier of the Gemini API. Google says Google AI Ultra subscribers can also try it in AI Studio and Google Antigravity, its agent-first development platform. The company describes the API and enterprise launch as global.[2][8]

Google positions Argon for long-horizon work that needs deep reasoning and structural precision, including code migrations and frontier research. It names software engineering, enterprise knowledge work and cybersecurity defence as target uses. Those are intended use cases, not independent proof that the model will improve a Malaysian firm’s output or reduce its costs.[2]

Google’s 13 October announcement sets an introductory API price of US$2 per million input tokens and US$10 per million output tokens. Cached input tokens receive a 95 per cent discount. After the introductory period ends on 30 November 2026, Google says the public model’s rates become US$4 per million input tokens and US$20 per million output tokens.[8]

These rates give developers a starting point for a budget, not a complete monthly cost. A team also needs to estimate input and output volume, cached-token use, currency conversion and any applicable tax. Google says Argon has a unified one-million-token input-and-output limit through the API, but that limit does not predict how many tokens a real workflow will consume.[8]

Google reports performance gains from internal and partner use. These remain company claims. Businesses should test their own tasks and keep human review before connecting a model to customer records or external actions.[8]