Gemini 4 Argon is open

Google opened Gemini 4 Argon on October 13. The company calls it a frontier model for long, difficult work in coding, finance, legal, cybersecurity, and science.

Who can use it is specific. Developers and enterprises get it through the paid Gemini API, in AI Studio and on the Gemini platform. Google AI Ultra subscribers get it in Google Antigravity, which Google describes as its agent-first development platform. Ultra and Pro subscribers get it in the Gemini app and in AI Mode in Search. Scientists can apply for the highest Antigravity usage tier, free for six months, onboarded as capacity allows. Cybersecurity teams can apply to the Fairwind program. Google says that program is beginning to expand to hundreds of additional partners, and that it offers Argon without cyber guardrails.

The price is an introduction. Google lists $2 per million input tokens and $10 per million output tokens, with cached input at 95 percent off. After that introductory period expires on November 30, 2026, Google says the price becomes $4 and $20, and that the higher price applies to the Argon model intended for public access. A budget built on the lower figures is a budget for a promotion.

The context window in the launch post is a 1 million token limit that covers input and output together, through the API. Google points developers to a continuation protocol in the documentation.

The quality claims that follow are Google’s, from the launch post.

Google says Argon ranks first on the Vals Index, which it describes as covering finance, coding, legal, and tax work. Inside Google, the company says Argon in Antigravity raised agentic code submissions by more than 35 percent and cut code rollbacks by nearly a third. It describes a Chrome refactor that separated core webpage handling from a legacy folder of about 6 million lines, inside a Chromium codebase it puts at about 40 million lines, and says months of architectural work finished in weeks. On the open-source Agent Development Kit, Google says an overnight run cut in-memory session preparation by 40 percent in long sessions and sped cold imports by 21 percent, and that the changes are in google/adk-python.

On safety, also in Google’s words: shortcut attempts on coding benchmarks fell to 2 percent, which it calls its best figure against earlier Gemini models, and attack success on the Gray Swan indirect prompt-injection benchmark fell to 0.7 percent across 15 adaptive attempts. The launch post points to a model card for the longer account.

The rest of the Gemini 4 family is not in this release. Google’s line is that Argon is the beginning.

If you already pay for the API, or for AI Pro or Ultra, the model is there to try. The introductory price expires on November 30.