Google opens access to flagship AI model Gemini 4 Argon

Google has opened access to its flagship neural network Gemini 4 Argon, which was previously in closed testing. The model is now available to users with Google AI Pro and Ultra subscriptions in the Gemini app and in the AI mode of the search system, as well as to developers through the Gemini API, Antigravity, and other proprietary services. According to 3DNews, citing 9to5Google, a subscription of the Ultra level is required to use Argon in the separate Google Antigravity application, its console version, as well as in Google AI Studio and the Android Studio Canary build.

The neural network is designed to solve tasks in the field of programming, cybersecurity, science, finance, and law. The API supports up to 1 million tokens (according to Habr, this is a single cumulative limit for input and output). According to Google, Argon showed a result of 68.9% in the Vals Index of professional tasks, 19.6% in the Harvey's Legal Agent Benchmark test, and 87% in the ERQA spatial thinking test.

Until November 30, the cost of access through the API is $2 for 1 million incoming tokens and $10 for 1 million outgoing tokens. From December 1, the rates will rise to $4 and $20, respectively. A 95% discount is available for cached incoming data.