Gemini 4 Argon Opens Up: Paid API Plus AI Pro and Ultra Users, $2/$10 Introductory Pricing Only Until November 30, and Not on Top of Third-Party Leaderboards

On October 13 Google widened Gemini 4 Argon from the Fairwind Program for cybersecurity defenders to broad availability: developers and enterprises can use it through the paid Gemini API, Google AI Studio and Antigravity, eligible Gemini Enterprise subscriptions include it, and Google AI Pro and Ultra subscribers get it in the Gemini app and AI Mode in Search, with individual Antigravity use requiring Ultra. Introductory API pricing is $2 per million input tokens and $10 per million output tokens, with cached input at $0.10, through November 30; from December 1 it returns to $4 and $20. The Fairwind Program also expands to hundreds of approved cybersecurity partners. Third-party results are mixed: on Vals AI's index as of October 12 Argon ranks first at 68.90%; on Artificial Analysis's Intelligence Index it scores 53, tied with GPT-6 Astra and Claude Fable 5.1 and below Claude Opus 5.5's 58.

Who can use it now

When it launched on September 30, Gemini 4 Argon was limited to vetted cybersecurity defenders in the Fairwind Program, a limited release this site covered at the time. Thirteen days later, on October 13, Google opened it up broadly:

  • Developers and enterprises: the paid Gemini API, Google AI Studio, Antigravity, and eligible Gemini Enterprise subscriptions; Android developers can use it through Android Studio and Firebase AI Logic;
  • Individuals: Google AI Pro and Ultra subscribers get it in the Gemini app and AI Mode in Search; individual use in Antigravity requires the Ultra tier.

The Fairwind Program also expands to hundreds of newly approved cybersecurity partners, with versions for authorized defensive work that drop some cyber restrictions.

The price has an end date

Introductory API pricing is $2 per million input tokens and $10 per million output tokens, with cached input at $0.10 after the 95% discount. That expires November 30; from December 1 it doubles to $4 and $20, matching Claude Opus 5.5. According to VentureBeat, the API supports up to 1 million combined input and output tokens through a continuation protocol, but Google hasn't clarified how that relates to the 1-million-token single-output ceiling announced in September.

What third parties say

Independent results disagree. On Vals AI's index as of October 12, Argon ranks first at 68.90%, ahead of Claude Sonnet 5.5 at 67.04% and Claude Opus 5.5 at 66.97%, and also tops its finance agent test. On Artificial Analysis's Intelligence Index it scores 53, tied with GPT-6 Astra and Claude Fable 5.1, below Claude Opus 5.5's 58 and the 56 of both Claude Sonnet 5.5 and Meta Muse Wonder Full. Google's own earlier Terminal-Bench 4.0 result still trails Opus 5.5 by about 9 points.

If you plan to try it, the window before November 30 is the cheapest: run comparisons first on scenarios where it leads, such as long legal and financial documents and long-video analysis, and test terminal and coding agent tasks against Opus 5.5 and Sonnet 5.5 before deciding.

via: Google announcement, VentureBeat report