Google is now opening Gemini 4 Argon to paying Gemini subscribers, developers, and businesses. Initially presented to a select circle of cybersecurity players as part of the Fairwind program, the model is being rolled out worldwide, with access conditions varying by profile, and the Gemini app is evolving in the process.
A more direct model, built for long tasks
Google positions Gemini 4 Argon for complex tasks spanning many steps, in code, finance, law, cybersecurity, or science. For the general public, the company mainly highlights a change in style. According to Google, Argon “answers questions more directly, with fewer precautions,” and the model should also produce fewer repetitions and clichés. On long tasks, it reports its progress more regularly.
For developers, Argon features in the Gemini API, i.e., the interface for integrating the model into software, a unified limit of one million tokens, shared between input and output. Tokens are the units of text the model reads and produces. Google accompanies this limit with a new “continuation protocol,” described in its documentation. For agentic uses, where the model acts autonomously on systems, the company indicates it has trained it to respect the limits of the systems it accesses and to honestly report negative results. Gemini 4 Argon is also meant to pause to obtain human confirmation before any irreversible action.
Access reserved for paid plans
Google AI Pro and Ultra subscribers use Argon in the Gemini app and in Google Search’s AI Mode, while Google AI Ultra subscribers can also use it in Antigravity. Developers access it via the paid Gemini API offering, in AI Studio and Android Studio. Businesses find it in Gemini for business, in the API, and in Antigravity. Scientists can apply to get free access for six months to the highest usage tier of Antigravity, with Google planning for 10,000 spots. The Fairwind program, which provides access to Argon without cybersecurity safeguards, is opening to hundreds of additional partners. Google does not mention Google AI Plus subscribers or free version users.
Regarding pricing, Google is applying a launch price until November 30, 2026, of $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached input tokens. The price will then increase to $4 for input and $20 for output.
Antigravity integrated into the Gemini app on Mac
Google is also integrating Antigravity, its agentic development platform, directly into the Gemini app on macOS. This makes it possible to switch from conversation to writing code without changing tools, an integration reserved for Google AI Ultra subscribers. Gemini’s main window also offers a toggle between two modes, Chat and Tasks, similar to its competitors. The second, which replaces Gemini Spark, allows users to assign the assistant work that it continues even when the user is not at their computer. Finally, Gemini connects to over 40 apps, including Instacart, OpenTable, Adobe, and Shopify.
