The key facts

  • Gemini 4 Argon was introduced in late September, broad access opened on October 13
  • API prices: $2/$10 per million tokens, from December 1 — $4/$20
  • The model was used to refactor 6 million lines of Chrome code and to speed up Gemma nearly threefold

Broad access to Gemini 4 Argon

Google announced the opening of broad access to its flagship artificial intelligence model Gemini 4 Argon. The model was first introduced in late September, when it was available only to selected testers and cybersecurity specialists as part of trials.

Where the model is available and how much it costs

From October 13, access to Argon is available to users with a Google AI Pro or Ultra plan in the Gemini app and the AI mode of the search engine, via API in AI Studio, as well as to Ultra subscribers in Google Antigravity. Using the model via API initially costs $2 per million input tokens and $10 per million output tokens. From December 1, Google will raise prices to $4 and $20 respectively.

Stated areas of application

According to the developers, the model is already being used in a number of areas: solving complex programming tasks, including optimizing application code; optimizing processes in the legal, financial, scientific and other industries; finding and fixing critical vulnerabilities in cybersecurity; and advanced research in quantum computing and applied science.

Examples of use

Google reported that with the help of Argon, Chrome browser developers completed a refactoring of files in a few weeks instead of months, isolating elements important for web page processing from 6 million lines of legacy code. The company also stated that Argon helped speed up the operation of the open Gemma models nearly threefold while preserving quality.